Retrospective Theses and Dissertations
1996
The structure and function of human brain
hexokinase
Chenbo Zeng
Iowa State University
Follow this and additional works at: http://lib.dr.iastate.edu/rtd
Part of the Molecular Biology Commons
Recommended Citation
Zeng, Chenbo, "The structure and function of human brain hexokinase " (1996). Retrospective Theses and Dissertations. Paper 11507.
This Dissertation is brought to you for free and open access by Digital Repository @ Iowa State University. It has been accepted for inclusion in
Retrospective Theses and Dissertations by an authorized administrator of Digital Repository @ Iowa State University. For more information, please
contact [email protected].
INFOKMAHUN TO USERS
This manuscript iias been reproduced from the microfilm master. UMI
films the text directly &om the original or copy submitted. Thus, some
thesis and dissertation copies are in ^pewriter face, while others may
be from aiqr type of conqniter printer.
Tbe quality of this reprodvction is dqtendoit upon the quality of the
copy submitted. Broken or indistinct print, colored or poor quality
illustrations and photographs, print bleedthroi^h, substandard margins,
and in^oper aligmnent can adverse^ affect reproduction.
In the unlikely event that the author did not send UMI a complete
manuscript and there are missing pages, these will be noted. Also, if
unauthorized copyright material had to be Temoved, a note wiDindicate
the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and
continuing from left to right in equal sections with small overly. Each
original is also photographed in one exposure and is included in
reduced form at the back of the book.
Photogrq>hs inchided in the original manuscr^t have been reproduced
xerographically in this copy. Higher quality 6" x 9" black and white
photographic prints are available for sssy photographs or illustrations
appearing in this copy for an additional charge. Contact UMI direct^
to order.
A Bell & Hcjwell Information Company
300 North Zeeb Road, Ann Arbor. Ml 48106-1346 USA
313/761-4700 800/521-0600
.w
The structure and function of human brain hexokinase
by
Chenbo Zeng
A dissertation submitted to the graduate faculty
in partial fulfillment of the requirements for the degree of
DOCTOR OF PHILOSOPHY
Department: Biochemistry and Biophysics
Major: Biochemistry
Major Professor; Herbert J. Fromm
Iowa State University
Ames, Iowa
1996
UMI Number: 9635373
UMI Microform 9635373
Copyright 1996, by UMI Company. All rights reserved.
This microform edition is protected against unauthorized
copying under Title 17, United States Code.
UMI
300 North Zeeb Road
Ann Arbor, MI 48103
ii
Graduate College
Iowa State University
This is to certify that the doctoral dissertation of
Chenbo Zeng
has met the dissertation requirements of
Iowa State University
Signature was redacted for privacy.
Major Mofessor
Signature was redacted for privacy.
For the Major Department
Signature was redacted for privacy.
For/the Graduate College
iii
TABLE OF CONTENTS
CHAPTER 1. GENERAL INTRODUCTION
Introduction
Dissertation Organization
References
CHAPTER 2. ACTIVE SITE RESIDUES OF HUMAN BRAIN HEXOKINASE
AS STUDIED BY SITE-SPECIFIC MUTAGENESIS
Abstract
Introduction
Experimental Procedures
Results
Discussion
References
CHAPTER 3. THE ATP-BINDING SITE OF HUMAN BRAIN HEXOKINASE
AS STUDIED BY MOLECULAR MODELING AND
SITE-DIRECTED MUTAGENESIS
Abstract
Introduction
Experimental Procedures
Results and Discussion
Acknowledgments
References
1
5
6
9
9
10
14
18
25
29
34
34
35
38
44
61
61
CHAPTER 4. A STUDY OF THE ATP-BINDING SITE OF HUMAN
BRAIN HEXOKINASE
Abstract
Introduction
Experimental Procedures
Results
Discussion
References
65
CHAPTER 5. CRYSTALLIZATION AND PRELIMINARY X-RAY
ANALYSIS OF HUMAN BRAIN HEXOKINASE
Abstract
Contents
Acknowledgments
References
84
CHAPTER 6. GENERAL CONCLUSIONS
92
ACKNOWLEDGMENTS
96
65
66
68
71
76
81
83
84
87
88
1
CHAPTER 1. GENERAL INTRODUCTION
Introduction
Brain hexokinase, the first enzyme of the glycolytic
pathway, catalyzes the phosphorylation of glucose using ATP
as the phosphoryl donor (1,2).
It is a crucial enzyme in the
regulation of energy metabolism in the brain (1,2).
The
molecular mass of human brain hexokinase is about 100,000
daltons, twice that of yeast hexokinase (3-7).
The N- and C-
terminal halves are homologous to each other and to yeast
hexokinase in the primary sequence (3-7), probably as a
result of gene duplication and fusion.
In 1989 White and
Wilson (8) reported studies on the selective tryptic cleavage
of rat hexokinase I into discrete N- and C-terminal halves.
They found that the active site of the enzyme resides in the
C-terminal half of hexokinase I.
Their kinetic results
suggest that the C-terminal portion of brain hexokinase,
mini-hexokinase, is virtually identical to the intact enzyme.
Recently, Magnani et al. (9) and our laboratory (10) obtained
mini-hexokinase from human tissue by recombinant technology
and, like White and Wilson (8), found that the truncated
enzyme is very similar in kinetic properties to hximan
hexokinase I.
The kinetic mechanism of brain hexokinase is thought to
be random Bi Bi (11-13), and a mechanism for its molecular
regulation has been proposed (14,15). This enzyme is strongly
2
inhibited by glucose-6-P and this inhibition can be relieved
by inorganic phosphate (Pi).
This has been an attractive
model for regulation of glycolysis in the brain.
The de-
inhibition of hexokinase and stimulation of
phophofructokinase by .Pi would cause .increased production and
removal of glucose-6-P simultaneously, resulting in an
increased glycolytic flux.
The site of action glucose-6-P inhibition has been a
subject of controversy for many years ever since 1954 when
Crane and Sols (16) made their proposal for a separate site
(other than the active site) on the enzyme.
This proposal
was based on the failure of glucose to reverse the inhibition
and on the kinetically observed differences in the
specificities for the binding of the glycosyl portions of
glucose and glucose-6-P to the enzyme.
Another proposal for
an allosteric site was also attractive from an evolutionary
view point.
It has been suggested that the brain enzyme
might have evolved by duplication and fusion of a gene coding
for yeast hexokinase and by subsequent evolution of one of
the two catalytic sites into a regulatory one (17,18).
This
model predicts the existence of two binding sites for
glucose-6-P (active site and regulatory site) per molecule of
the enzyme.
However, in several binding studies, only one
binding site for glucose-6-P has been detected (19-21).
The
finding of Fromm and coworkers, that glucose-6-P inhibition
is competitive versus ATP (22), suggests that glucose-6-P
3
competes with the
y-phosphate group of ATP at the nucleotide
site on the enzyme.
Further, they suggested that the hexose
moiety of glucose-6-P might occupy an adjacent site within
the active-site domain but outside its glucose subsite.
The interaction of glucose-6-P and Pi has been
studied by a number of methods.
Kinetic studies on the
Hexokinase I from a variety of sources have indicated that at
physiological concentrations of Pi (3-5mM) Pi neither
activates the uninhibited enzyme nor does it alter the K„ for
ATP (23,24), however, it can relieve the inhibition by
glucose-6-P.
Direct binding studies of Pi to hexokinase
suggests that l mole of Pi binds per mol of enzyme.
The
binding of Pi at low concentrations does not appear to be
influenced by MgATP^"
and glucose-6-P
or glucose.
On the other hand MgATP^"
exhibit competitive binding (25). It seems
that Pi and ATP function by different mechanisms in
dissociating glucose-6-P
from hexokinase.
Those results
suggest the Pi binding site is distinct from the active site.
Wilson and his coworkers (26) have demonstrated that Pi binds
to the N-terminal half of the enzyme on the basis of its
ability to protect the N-terminal half against denaturation
and proteolysis.
The result from our lab supports this
conclusion (10).
The truncated hexokinase, mini-hexokinase,
starting with Met-455 and ending at the C-terminus lost its
ability to ameliorate inhibition of glucose-6-P-inhibited
mini-hexokinase in the presence of phosphate.
We suggest
4
that the Pi site either resides in the N-terminal half of
hexokinase I or requires the N-terminal portion of the
enzyme.
Recently, Tsai and Wilson (27) constructed the
chimeric hexokinase from the N- and C-terminal domains of the
rat type I and type II isozymes.
The chimeric hexokinase
consists of either the N-terminal half of type I hexokinase
fused with the C-terminal half of the type II isozyme or the
inverse pair.
The response of the chimeras to Pi was
correlated with the origin of the N-terminal domain.
finding
This
also supports the conclusion that the Pi binding
site is in i:he N-terminal portion.
The structure of yeast hexokinase has long been
available (28-30), and has been used as a model
specific mutagenesis studies of hexokinase.
for site-
The glucose-
binding domain has been proposed from these investigations.
Unfortunately,
the crystal structure of the yeast enzyme
provides no information on the ATP binding domain within the
active site. In 1992, Bork et al. and others (31-33)
suggested that the heat shock protein hsc70, actin, and yeast
hexokinase may have similar three-dimensional structures in
their ATP-binding sites.
The ATP-binding motif is distinctly
different from the single phosphate-binding loop in, for
example, adenylate kinase, rec A protein, or ras p21 (34).
Amino acid residues in the ATP-binding site are conserved
although no overall sequence similarity among these three
proteins is found.
Those conserved residues are also found
5
in human brain hexokinase.
The major work of this dissertation is to establish the
ATP-binding site by molecular modelling and site-directed
mutagenesis.
An effort was devoted to the crystallization of
human brain hexokinase.
The Pi binding site was also
studied.
Dissertation Organization
This dissertation is organized into six chapters.
Chapter 1 provides the background of study, statement of the
problems and objectives of the research project.
Chapter 2
to 5 present four papers that are either published or
submitted to the scholarly journals.
Chapter 2 reports
construction of a vector to overexpress the truncated
hexokinase, also studies the inorganic phosphate binding site
and ATP binding site.
Chapter 3 is devoted to the
establishment of the -ATP-binding site model of human brain
hexokinase.
Chapter 4 further studies amino acid residues in
the ATP binding site, especially the role of glycine residues
in the catalytic reaction of human brain hexokinase.
Chapter
5 describes the crystallization of human brain hexokinase in
order to resolve the three-dimensional structure of human
brain hexokinase.
Chapter 6 includes the concluding remarks
and recommendations.
References
1.
Lowry, O.H., and Passonneau, J.V. (1964) J. Biol. Chem.
239, 31-42.
2.
Rapoport, S. (1968) Essays Biochem. 4, 69-103.
3.
Easterby, J. S., and O'Brien, M. J. (1973) Eur. J.
Biochem. 38, 201-211-
4.
Rose, I. A., Warms, J. V. B., and Kosow, D. P. (1974)
Arch. Biochem. Biophvs. 164,729-735.
5.
Holroyde, M. J., and Trayer, I. P. (1976) FEBS Lett. 62,
215-219.
6.
Ureta, T. (1982) Comp. Biochem. Physiol. 71B, 549-555.
7.
Manning, T. A., and Wilson, J. E. (1984) Biochem.
Biophys. Res. Commun. 118, 90-96.
8.
White, T. K., and Wilson, J. E. (1989) Arch Biochem.
Biophys. 274, 375-393.
9.
Magnani, M., Bianchi, M., Casabianca, A., Stocchi, V.,
Daniele, A., Altruda, F., Ferrone, M. and Silengo, L.
(1992) Biochem. J. 285, 193-199.
10. Zeng, C., and Fromm. H. J. (1995) J. Biol. Chem. 270,
10509-10513
11. Ning, J., Purich, D. L., and Fromm, H. J. (1969)
J.
Biol. Chem. 244, 3840-3846
12. Bachelard, H. S., Clark, A. G., and Thompson, M.
F.(1971) Biochem. J. 123, 707-715
13. Gerber, G., Preissler H., Heinrich, R., and Rapoport, S.
M. (1974) Eur. J. Biochem. 45, 39-52
14. Fromm, H.J. (1981) in The Regulation of Carbohydrate
Formation and Utilization in Mammals (C. M. Veneziale
Ed.), University Park Press, Baltimore, pp45-68.
15. Wilson, J. E. (1984) in Regulation of Carbohydrate
Metabolism (Beitner, R., ed)
Vol. 1, pp. 45-85, CRC
Press, Inc., Boca Raton, PL.
16. Crane, R.K., and Sols, A. (1954) J. Biol. Chem. 210,
597-606.
17. Easterby, J. S. (1971) FEBS Lett. 18, 23-26
18. Colowick, S. P.(1973) in The enzvmes. 3rd edn(Boyer, P.
D., ed) vol. 9, pp. 1-48, Academic Press, New York.
19. Ellison, W. R., Lueck, J. D., and Fromm, H. J.(1975) J.
Biol. Chem. 250, 1864-1871
20. Mehta, A., Jarori, G. K., and Kenkare, U. W. (1988) J.
Biol. Chem. 263, 15492-15497.
21. Chou, A. C. and Wilson, J. E. (1974) Arch. Biochem.
Biophys. 165, 628-633
22. Fromm, H. J. and Zewe, V. (1962) J. Biol. Chem. 237,
1661-1667.
23. Kosow, D. P., Lski, F. A.k Warms, J. V. B., and Rose, I.
A.(1973) Arch. Biochem. Biophys. 574, 114
24. Purich, D. L. and Fromm, H. J.(1971) J. Biol. Chem. 246,
3456-3463.
25. Ellison, W. R., Lueck, J. D. and Fromm, H. J. (1974)
Biochem. Biophvs. Res. Commun. 57, 1214-1220.
26. White, T. K., and Wilson, J. E.(1990) Arch Biochem.
Biophys. 277, 26-34
27. Tsai, H. J., and
Wilson, J. E. (1995) Arch Biochem.
Biophys. 316, 206-214
28. Anderson, C. M., Stenkamp, R. E., McDonald, R. C., and
Steitz, T. A. (1978) J. Mol. Biol. 123, 207-219.
29. Bennett, W. S., Jr., and Steitz, T. A. (1980) J. Mol.
Biol. 140, 211-230.
30. Harrison, R. (1985) Crystalloaraphic Refinement of Two
Isozymes of Yeast Hexokinase and Relationship of
Structure to Function. Ph. D. Thesis, Yale University,
New Haven, CT.
31. Bork, P., Sander, C. and Valencia, A. (1992) Proc. Natl
Acad. Sci. USA 89, 7290-7294.
32. Kabsch, W., Mannherz, H. G., Suck, D., Pai, E. F. and
Holmes, K. C. (1990) Nature 347, 37-44.
33. Flaherty, K. M., McKay, D.B., Kabsch, W. and Holmes, K.
C. riggit Proc. Natl. Acad. Sci. USA 88, 5041-5045.
34. Schulz, G. E. (1992) Curr. Qpin. Struct. Biol. 2, 61-67
9
CHAPTER 2. ACTIVE SITE RESIDUES OF HUMAN BRAIN HEXOKINASE AS
STUDIED BY SITE-SPECIFIC MUTAGENESIS^
A paper published in Journal of Biological Chemistry^
Chenbo Zeng^ and Herbert J.Fronim^'"
Abstract
The truncated gene of hexokinase, mini-hexokinase,
starting with methionine*®® and ending at the C-terminus was
expressed in Escherichia coli.
Mini-hexokinasae lost its
ability to ameliorate inhibition of glucose-6-P inhibited
mini-hexokinase in the presence of phosphate (Pi).
We
suggest that the Pi site either resides in the N-terminal
half of hexokinase I or requires the N-terminal portion of
the enzyme.
Site-directed mutagenesis was performed to
obtain two mutants of mini-hexokinase: C606S, C628S.
Both
are thought to be associated with the active site of
^This research was supported in part by Research Grants
NS10546 from the National Institute of Health, United State
Public Health Service, and Grant MCB-9218763 from the National
Science Foundation. This is Journal Paper J-16213 of the Iowa
Agriculture and Home Economics Experiment Station, Ames,
Project 2575.
^Reprinted
10509-10513.
with
permission
of J. Biol. Chem.. 1995,
270,
^Graduate student and Professor, respectively. Department of
Biochemistry and Biophysics, Iowa State University.
Research
conducted and manuscript written by Zeng with assistance from
Fromm.
•*To whom correspondence should be addressed.
10
hexokinase I. These mutants exhibited a 3-fold increase in Km
for glucose, but no cliange in either the Km for ATP or the
k„t*
The circular dichroism spectra (CD) showed no
differences among the wild-type or mutant enzymes.
These
results suggest that Cys606 and Cys628 are not involved in
glucose binding directly.
The putative ATP binding site of
full-length human brain hexokinase may involve R539 and G679,
and these residues were mutated to lie. For the mutant R539I,
the k„tvalue decreased 114-fold relative to wild-type
hexokinase, whereas the K, values for ATP and glucose changed
only slightly.
No ctaange was observed for the Ki value for
l,5-Anhydroglucitol-6-phosphate. CD spectra showed only a
slight change in secondary structure. For the mutant G679I,
overexpressed hexokinase is insoluble. We suggest R539 is
important for catalysis by stablizing the transition state
product ADP-hexokinase. G679 is probably important for proper
folding of the protein.
Introduction
Brain hexokinase,
hexokinase I (ATP:D-hexose 6-
phosphotransferase,EC 2.7.1.1), is thought to be the
pacemaker of glycolysis in brain tissue (1) and the
erythrocyte (2).
The cDNAs for hexokinases from a variety of
tissues, including yeast (3,4), human kidney (5), rat brain
(6), rat skeletal muscle (7), rat liver (8,9), and mouse
hepatoma (10) have been cloned, and their complete amino acid
11
sequences have been deduced from the cloned cDNAs.
The C-
terminal domains of mammalian hexokinases and their Nterminal domains have overall similarities to each other and
to yeast hexokinase, probably as a result of gene duplication
and fusion (11-15).
The crystal structure of yeast
hexokinase is available (16,17), and the putative glucose
binding domain has been suggested from these investigations
(8,18).
The involvement of a cysteinyl residue either in or near
the active site of brain hexokinase (19-22) has been
suggested.
Swarup and Kendare were the first investigators
to use N-(bromoacetyl)-D-glucosamine to covalently modify
brain hexokinase (23). They demonstrated that two sulfhydryl
residues are involved in covalent modification with
concomitant enzyme inactivation; however,the enzyme could be
protected against inactivation by either ATP or glucose.
Recently, Schirch and Wilson (24) used this same protocol
with rat brain hexokinase and obtained results similar to
those reported by Swarup and Kenkare (23), and concluded that
there was a single critical sulfhydryl based on the
observation that modification of that sulfhydryl totally
inactivated the enzyme. Later, in a subsequent study, they
isolated and partially sequenced two peptides that are
modified by N-(bromoacetyl)-D-glucosamine and that are
protected against modification by either glucose or Nacetylglucosamine (25).
By locating these two peptides
12
within the tertiary structure of yeast hexokinase, based on
the structure proposed by Steitz and coworkers (16,17), they
found that these peptides are close to the binding site for
glucose.
Comparing the amino acid sequences between yeast
hexokinase and the brain hexokinase peptides, they found no
cystine residues corresponding to the peptides in question.
Based on the proximity of the peptides to the glucose binding
site and the analysis of sequence homology between yeast
hexokinase and those peptides, they questioned the existence
of a single critical sulfhydryl.
To solve this problem
directly, we mutated Cys606 and Cys628, which are located in
those two peptides.
Working with adenylosuccinate synthetase
(26), we found a thiol residue to be essential when
covalently modified with 5,5'-dithiobis (2-nitrobenzoic
acid); however, site-specific mutagenesis of the specific
cysteinyl residue in question revealed that the covalent
modification caused inactivation through a steric effect,
i.e., the enzyme could not fold into its native conformation
after modification v/ith 5,5'-dithiobis (2-nitrobenzoic acid).
In 1989 White and Wilson (27) reported studies on the
selective tryptic cleavage of rat hexokinase I into discrete
N- and C- terminal halves.
They found that the active site
of the enzyme resides in the C-terminal half of hexokinase I,
and they suggested that the allosteric site for glucose-6-P
is contained in the N-terminal segment of brain hexokinase.
Their kinetic results suggest that the C-terminal portion of
13
brain hexokinase,
mini-hexokinase,is virtually identical to
the intact enzyme. Recently, Magnani et al. (28) obtained
mini-hexokinase from human tissue by recombinant technology
and, like White and Wilson (27),
found that the truncated
enzyme is very similar in kinetic properties to human
hexokinase I.
In the present study, we report the functional
expression of human brain mini-hexokinase
in Escherichia
coli and its purification to homogeneity.
Site-directed
mutagenesis was performed to obtain two mutants of minihexokinase, C606S and
C628S, in order to evaluate the roles
of these two cysteinyl residues in the hexokinase reaction.
Although our laboratory has cloned and expressed full-length
human brain hexokinase in E.coli. we chose to investigate
mini-hexokinase to simplify interpretation of the sitespecific mutagenesis data and, in addition, to study the
effects of inorganic phosphate (Pi) on glucose-6-P inhibited
mini-hexokinase.
Because the crystal structure of yeast hexokinase did
not contain ATP, the location of this substrate has not been
defined. On the other hand, in actin and hsp70 heat shock
protein there is a binding site for ATP, and it has been
suggested (29-31) that a similar ATP binding domain may exist
on hexokinase. To test this hypothesis, we used the
structures of hsp70 and actin as a model in an attempt to
locate the nucleotide binding site on human brain hexokinase
14
by using the technique of site-directed mutagenesis.
Experimental Procedures
Materials—The pET-lld plasmid was obtained from
Novagen. E.Coli strains BL21(DE3) and TG-1 cells were
obtained from Amersham Corporation. Ncol and SphI were from
New England Biolabs, and BamHI and T4 DNA ligase were from
Promega.
The
oligonucleotide-directed in vitro mutagenesis
system, version 2.1, was a product of Amersham Corporation.
The magic minipreps DNA purification system, was a product of
Promega.
Oligonucleotide synthesis and nucleotide sequence
analysis were done by the Iowa State University Nucleic Acid
Facility. Ampicillin, DL-dithiothreitol (DTT)®,
phenylmethylsulfonylfluride (PMSF), ATP, NADP, NADPH,
glucose-l,6-bisphosphate (Glc-l,6-BP), 1,5-anhydro-Dsorbitol, streptomycin, phosphoenolpyruvate,
deoxyribonuclease I (DNasel), lactatedehydrogenase /pyruvate
kinase, and lysozyme were products of Sigma. Isopropyl-6-Dthiogalactopyranoside (IPTG) was purchased from Clontech.
Bio-gel HT was obtained from Bio-Rad Laboratories.
Expression of mini-hexokinase pET-lld-mHKI—The full
length cDNA for the human brain hexokinase I has been cloned
into the expression vector pET-lla in our laboratory (32).
®The abbreviations used are; DTT, DL-Dithiothreitol; PMSF,
phenylmethyl-sulfonylfluoride; IPTG, Isopropyl-B-Dthiogalactopyranoside; Glu-1,6-BP, Glu-1,6-bisphosphate;
G6P,glucose-6-phosphate; DNase I, Deoxyribonuclease I.
15
This full length cDNA clone was further digested with
restriction enzyme Ncol and BamHI to obtain a 1399bp NcolBamHI fragment, which encodes a truncated form of hexokinase
starting from methionine-455 to the terminal amino acid.
The
NcoI-BamHI fragment was then purified and ligated into
plasmid pET-lld, containing the bacteriophage T7 promoter,
previously digested with Ncol and BamHI to generate the pETlld-mHKI clone.
pET-lld-mHKI was transformed into E. Coli
strain BL21 (DE3), a phage lambda lysogen containing the T7
RNA polymerase gene under control of the lacUVS promoter. A 5
ml pregrowth culture of the transformed strain in M9 medium,
plus 40mg/L ampicillin and 4mg/L glucose, was grown overnight
and then added to 500ml of the same medium.
The culture was
shaken at 37° C to early log phase (A6oo=0.4), and IPTG was
then added to a final concentration of 0.4mM to induce the T7
RNA polymerase gene. Cultures were shaken vigorously for an
additional 24 to 40 hours at 20° C.
Purification of mini-hexokinase and full-lenath
hexokinase—Protein purification was performed essentially as
described previously (32) with some modifications.
Cells
were harvested by centrifugation at 4000xg for 10 min and
washed once with lOmM KPi, pH7.0. The resultant pellet was
resuspended in 120ml 5mM KPi,pH7.0,containing ImM EDTA, 50mM
glucose, ImM DTT,9%
PMSF.
glycerol, lysozyme (2mg/mL), and ImM
The cell suspension was kept at 0° C for 1 hour and
then sonicated four times for 20s at 40W.
To the mixture.
16
600/il diluted DNAase (150/il lOmg/ial DNAase stock +650ul IM
MgCla) was added, and the suspension was incubated at room
temperature for 20min.
xg for 45min.
The lysate was centrifuged at 17500
The supernatant fluid was purified further by
streptomycin, (NH4)2S04
precipitation, heating, and an ion-
exchange HPLC column step as described elsewhere (32).
The
concentrated fractions with high enzyme activity were then
loaded onto a hydroxyapatite (HT) column (lx25cm) previously
equilibrated with 25mMK, pH7.0 ,containing ImM DTT.
A
linear 360-ml gradient from 25mM to 0.5M KPi was used to
eliate mini-hexokinase.
The fractions with high enzyme
activity were pooled, concentrated, and then injected onto
Bio-Sil SEC gel-filtration HPLC columns (600mm x 7.5mm and
300mm X 7.5mm connected together).
Hexokinase activity was
eluted with 50mM KP^ buffer, pH 7.0, containing O.IM KCl and
1 mM DTT.
Site-directed mutagenesis—Mutagenesis was performed by
following the instructions provided with the
oligonucleotide-directed in vitro mutagenesis system version
2.1 from Amersham. The oligonucleotides used for mutagenesis
are as follows;
5—ACC-AAT-TTC-ATT-GTG-CTG-CTG-3
for R539I,
5—CTC-ATT-GTT-ATT-ACC-GGC-AGC-3
for G679I,
5—TCA-TTT-CCC-TCC-CAG-CAG-ACG-3
for C606S,
5—GCA-ACA-GAC-TCC-GTG-GGC-CAC-3
for C628S,
where the underlines represent the mutated bases. The DNA
17
sequence of the mutated gene was verified by DNA sequencing
using the chain termination method (33).
After mutagenesis,
the BamHI-Sph I fragment of the mini-hexokinase gene was
cloned back into the expression vector pET-lld, similarly
digested with BamHI and SphI, and the BamHI-BamHI fragment of
full-length hexokinase was cloned back into the expression
vector pET-lla, which is also digested with BamHI, creating
pET-lld-C606S, pET-lld-C628S, pET-lla-R539I, and pET-llaG679I,
Hexokinase activity assay—Hexokinase activity was
determined spectrophotometrically as described previously
(32).
Preparation of 1.5-An-G6P—l,5-An-G6P was prepared as
described previously (34).
Determination of protein concentration—Protein
concentration was determined by the method of Bradford (35),
using bovine serum albumin as a standard.
Circular Dichroism Measurements—Circular Dichroism (CD)
spectra were recorded under the supervision of Dr. E.
Stellwagon at the University of Iowa.
Measurements were made
at room temperature in a 0.1 cm pathlength cylindrical quarts
cell.
The concentration of the sample was 0.15-0.66mg/ml in
2mM Hepes,PH7.0, containing 20mM KCl and 0.2mM
mercaptoethanol.
buffer solution.
Base lines were obtained using protein-free
18
Results
Expression.purification. and characterization of wildtvpe mini-hexokinase—E.coli strain BL21:DE3 (pET-mHKI)
expresses a 50KDa protein upon induction with IPTG.
Mini-
hexokinase was purified from induced E.coli containing pETlld-mHKI to near homogeneity as described in "Experimental
Procedures" using SDS-PAGE as the criterion (data not shown).
The specific activity of mini-hexokinase is 62, which is
close to the value for full-length hexokinase from human
brain tissue (32).
The K„ values of the substrates were
determined to be 26iiK and 2.9mM for glucose and ATP,
respectively.
Mini-hexokinase is inhibited competitively
relative to MgATP by glucose-l,6-bisphosphate (Fig. 2.1).
A
cardinal feature of glucose-6-P-inhibited hexokinase I is the
ability of Pi to reverse the inhibitory effects of the
product (36,37).
deinhibition by Pj
Experiments of glucose-6-P inhibition and
were carried out to determine the effect
of Pi on the glucose-6-P-inhibited mini-hexokinase.
The
results indicated that mini-hexokinase is sensitive to
inhibition by glucose-6-P, but 6 mM P^ cannot relieve the
inhibition caused by glucose-6-P. P^ at 6mM has little effect
on mini-hexokinase in the absence of glucose-6-P (Fig. 2.2B).
For full-length hexokinase I, Pi at 6mM does ameliorate
glucose-6-P inhibition (Fig. 2.2A)
(32,38).
Therefore, we
suggest that either the P^ site resides in the N-terminal
half of hexokinase I, or that the N-terminal half of
19
4.50-
3.60-
" 1.80-
0.90-
O.OO
0.00
0.70
2.10
1.40
(1/ATP)(1/mM)
280
3.50
7.00-
6.005.00.S- 4.00-
u
a>
^ 3.00-
2.001.000.00-
0.00
0.20
0.60
0.40
(1/ATP)(1/mM)
0.80
1.00
Fig. 2.1. Inhibition of hexokinase I by glucose-1,6bisphosphate. Fig. 2.1A, full length hexokinase I; Fig.
2.IB, mini-hexokinase. The concentration of glucose-1,6bisphosphate was OpiMC), 100/iM(+), 167/iM(A). Activity was
determined using the glucose-6-phosphate dehydrogenase
coupled assay. The measured change in absorbance per min was
plotted as percentage of the enzyme activity at saturating
concentrations of glucose and MgATP. The apparent Kj
values for both mini-hexokinase and full length hexokinase
I is lOO^M.
20
1.00-
0.80-
'S 0.60*
^ 0.40-
0.20-
0.00-
0:00
2.00
1;00
4.00
3.00
(1/ATP)(1/mM)
1.00-
C
B
o
5 0.50u
o
(D
5:
0.00-
0.00
0.20
0.40
0.80
1.00
Fig. 2.2. Effectiveness of inorganic phosphate on
reversing inhibition by l,5-An-G6P. Fig. 2.2A, full-length
hexokinase I; Fig. 2.2B, mini-hexokinase. Concentration of
glucose was 2 mM. Concentration of l,5-An-G6P and Pi were;
(•), 0, 0; (+), 50mM, 0; (A), 50 /iM, 6 inM, respectively.
21
hexokinase I is required for deinhibition of glucose-6-Pinhibited hexokinase I.
Site-directed mutations of Ara539|, G1y679 in the
putative ATP-bindina site—It has been reported that yeast
hexokinase, actin, and hsp70 heat shock proteins have similar
three-dimensional structures in their ATP binding sites, and
in addition, the amino acid residues are conserved in the ATP
binding sites (29-31). Based on this suggestion, R539 and
G679 of human brain hexokinase might be involved in ATP
binding. The R539 and G679 residues were mutated to lie by
site-directed mutagenesis. Two mutants, D539I and G679I, were
created by site-directed mutagenesis. The mutant D539I was
purified to homogeneity and characterized kinetically (Table
2.1).
The keat value decreased 114-fold,the K„ value for ATP
increased 2-fold, and the K„ value for glucose decreased 4.6fold. Little change was observed for the Ki values for 1,5An-G6P. For the mutant G679I, overexpressed hexokinase is
insoluble. SDS PAGE indicated that mutant G679I hexokinase
was overexpressed, but the cell extract did not contain
hexokinase (data not shown).
The G679 residue is probably
important for proper folding of the protein.
Site-directed mutations of Cys606. Cvs628 in the
putative alucose-bindina site—Two mutants of mini-hexokinase
were prepared to investigate whether Cys606 and Cys628 are
involved in glucose binding and/or catalysis.
Both mutants,
C606S and C628S, exhibit 3-fold increases in K„ values for
Table 2.1. A Comparison of the Kinetic data of the wild-type and the mutant forms of
hexokinase
Hexokinase
K^(Glu)
K„(ATP)
(/xM)
(mM)
K,(l,5-An-G6P)
(iM)
(s'^)
full-length wild-type
56 ±5.6
1.0±0.2
26 ±4.8
26.9
full-length Arg539Ile
12 ± 3.2
1.9 ± 0.3
49 ± 11
mini wild-type
26 ±4.0
2.9±0.3
22 ±2.5
mini Cys606Ser
87 ±9.1
2.2±0.4
18.6
mini Cys628Ser
89 ± 15
2.1 + 0.4
15.6
0.23
17.1
23
glucose compared with wild-type mini-hexokinase; however, no
alteration was observed on the K„ for ATP (Table 2.1).
In
addition, both mutants are similar to wild-type minihexokinase in enzyme activity.
The CD spectra for mutants
C606S and C528S and wild-type mini-hexokinase show no
differences (data not shown).
These results suggest that
Cys606 and Cys628 are not involved in catalysis and that
covalent modification of Cys606 and Cys628 by N(bromoacety1)-D-glucosamine very probably caused inactivation
by disrupting the native conformation of brain hexokinase.
Effect of mutations on the CD Spectra of Tninl-hexokinase
and full length hexokinase—To check whether the change in
K,
values of mutants is due to gross structural
perturbations, CD experiments were carried out for the wildtype enzyme, mini-hexokinase, and the mutants C606S, C628S,
and R539I.
C606S and C628S mutants exhibit no differences in
their CD spectra compared to that of the wild-type minihexokinase (data not shown).
These results suggest that the
mutant amino acid residues at Cys606
and Cys628
do not
cause changes in secondary structure. R539I and wild-type
hexokinase showed only a
slight difference in CD spectra
(Fig. 2.3).
Discussion
The finding by White and Wilson (27) that brain
hexokinase could
be cleaved into two roughly equal segments
i
24
10 T
-10 ^
>. -20 -
-40 ••
-50
200
220
240
260
wavelength(nm)
Fig. 2.3. CD spectra for wild-type hexokinase 1(B) and
mutant
D539I(A). The concentration of the sample was
adjusted to 660^g/ml.
25
of which the C- half is very similar, if not identical in
properties to the native enzyme is extremely interesting.
This observation is fully consistent with the suggestion that
hexokinase I arose through gene duplication and fusion (il­
ls).
Our studies with recombinant mini-hexokinase differ
from those of the Italian group (28) only in the system used
to express the enzyme.
When we tested glucose-6-P-inhibited
mini-hexokinase with levels of Pj known to deinhibit glucose6-P-inhibited full-length hexokinase (32,38), 6inM, no
deinhibition was observed.
We conclude, therefore, that the
allosteric Pj site proposed for brain hexokinase (39) either
resides in the N-terminal half of hexokinase I or requires
that portion of the molecule for expression. However, White
and Wilson (27) claimed that Pi ameliorates the inhibition of
mini-hexokinase by l,5-An-G6P. The reason that our results
are different from Wilson's findings is not clear. The
observation that mini-hexokinase exhibits a 3-fold increase
in K„ for ATP compared with the full-length enzyme suggests
that the truncated and native enzymes exhibit somewhat
different kinetic properties, e.g., the N-terminal region of
hexokinase I may be required for maintaining the proper
conformation of the active site.
In 1962 we found that glucose-6-P is a normal competitive
product inhibitor of ATP in the brain hexokinase reaction
(40).
This finding was clearly at variance with data from
the laboratory of Crane and Sols (41). It is now wildly
26
accepted that glucose-6-P
of ATP (28,40,42,43).
is indeed a competitive inhibitor
Although we suggested that glucose-6-P
is a normal product inhibitor of brain hexokinase,
others,notably Lazo et al. (44) and Wilson (45), have
maintained
that glucose-6-P binds at an allosteric site on
hexokinase I.
Reports from our laboratory, based primarily
on kinetic studies (40),
as well as NMR experiments from
other laboratories (46,47), are clearly at variance with the
notion that a topologically distinct site for glucose-6-P
exists on hexokinase I.
In addition, initial rate studies
from Sols laljoratory (48) as well as ours (49) on the back
reaction clearly eliminate the existence of a high affinity
glucose-6-P inhibitory allosteric site.
The finding by White
and Wilson (27) that mini-hexokinase is as susceptible to
glucose-6-P inhibition, both
quantitatively and
qualitatively, as the native brain enzyme is fully consistent
with the proposition that an allosteric site for the sugar
phosphate product
does not exist.
What is clear from the
present report is that the allosteric site for Pi on
hexokinase I is in some way associated with the N-terminal
half of the enzyme.
In 1992, Bork et al.and others (29-31) indicated that
the heat shock protein hsp70, actin, and yeast hexokinase
have similar three- dimensional structures in their ATP
binding sites. The ATP binding site consists of five motifs:
PHOSPHATE 1, CONNECT 1, PHOSPHATE 2, ADENOSINE, and CONNECT
27
2. The ATP binding motif is distinctly different from the
single phosphate binding loop in, for example, adenylate
kinase, rec A protein, or ras p21 (50). Amino acid residues
in the ATP binding site are conserved although no overall
sequence similarity among these three proteins is found. In
particular, the PHOSPHATE 1 motif, which interacts with the
6- and yphosphates of the nucleotide, shows sequence
similarity. In this motif, positively charged lysine or
arginine residues form a salt bridge to the negataively
charged nucleotide phosphates. In brain hexokinase, the
corresponding residue is Arg539. In this report Arg539 was
mutated to lie. The k<.at value for the mutant decreased 114fold relative to the wild-type enzyme. The K„ value for ATP
increased 2-fold, whereas the K. value for glucose decreased
4.6-fold. No change was observed for the Ki value for 1,5-AnG6P. CD spectra showed only a minor change in secondary
structure.
We suggest therefore that Arg539 is important
for catalysis. Since the currently available yeast hexokinase
structure (51) does not provide for an ATP site, we used
actin as a model for the ATP binding site. In actin the
residue corresponding to Arg539 is Lysl8 (29-31). The e-amino
group interacts with the side chain carboxylate group of
Aspll, which binds to the metal ion. In the ADP-actin
complex, the e-amino group forms hydrogen bonds with the aand B-phosphates. We suggest by analogy to actin that Arg539
is important for catalysis by stabilizing the transition
28
state product ADP-hexokinase complex. In the motif PHOSPHATE
2, instead of the conserved Aspl54 of actin/hsc70 in which
Aspl54 binds the metal ion, brain hexokinase has Gly679. When
Gly679 was mutated to lie, the overexpressed mutant, G679I,
was found to be insoluble. It is possible that G679I is
overexpressed in the form of inclusion bodies. We suggest
that G679 is probably important for the proper folding of the
protein.
A number of studies have shown that there is an
essential thiol group at the active site of hexokinase I (2224).
Schrich and Wilson (25) isolated three peptides
containing modified sulfhydryl groups after trypsin digestion
of the enzyme.
They found that labelling of two peptides,
one containing Cys606 and the other, Cys628, was sensitive to
the addition of protective ligands, glucose or Nacetylglucosamine.
One explanation for this observation is
that sulfhydryl groups are important for activity, contrary
to the previous conclusion that only one sulfhydryl group is
critical.
Another explanation is that neither sulfhydryl
group is important; however, modification of either
sulfhydryl by a large molecule, e.g., GlcNBrAc, will hinder
the binding of glucose.
To determine the importance of
Cys606 and Cys628, Cys residues were mutated to Ser at these
two positions.
Both of these mutants exhibited 3-fold
increase in K„ values for glucose, but no change in either
the K„ for ATP or the k^at was observed.
These results
29
suggested that Cys606 and Cys628
are not critical residues
for either catalysis or binding of substrates.
Amino acid
sequence comparisons between yeast hexokinases A and B and
brain hexokinase type I show no corresponding Cys residues
in yeast hexokinase, which also implies that these two Cys
residues might not be important in the functioning of
hexokinase I.
According to X-ray diffraction studies of
yeast hexokinase (16,17), Cys606 and Cys628 are located at
or near the glucose-binding site.
Their modification with
big molecules may hinder the entrance of the substrate and
lead to the loss of activity. Results of the current
investigation suggest that the X-ray diffraction
crystallographic structure of the ATP binding sites of actin
and hsp70 (29-31) are excellent models for the ATP binding
site of human brain hexokinase based upon site-specific
mutagenesis and kinetic studies.
References
1. Lowry, O.H., and Passonneau, J.V. (1964) J. Biol. Chem.
239, 31-42.
2. Rapoport, S. (1968) Essays Biochem. 4, 69-103.
3. Frohlich,K.-U., Entian,K.-D., and Mecke,D. (1985) Gene
CAmst.) 36,105-111.
4. Kopetzki, E., Entian, K.-D., and Mecke, D. (1985) Gene
fAmst.
39, 95-102.
30
5. Nishi, S., Seino, S., and Bell, G. E. (1988) Biochem.
Biophvs. Res. Cominun. 157,937-943.
6. Schwab,D.A., and Wilson, J.E. (1989) Proc. Natl. Acad.
Sci. U.S.A. 86, 2563-2567.
7. Thelen, A.P., and Wilson, J.E. (1991)
Arch. Biochem.
Biophys. 286,645-651.
8. Andreone, T. L., Printz, R. L., Pilkis, S. J., Magnuson,
M. A., and Granner, D. K. (1989) J. Biol. Chem. 264,
363-369.
9. Schwab, D. A., and Wilson, J. E. (1991) Arch. Biochem.
Biochvs. 285, 365-370.
10. Arora, K. K., Fanciulli, M., and Pedersen, P.L. (1990)
J. Biol. Chem. 265, 6481-6488.
11. Easterby, J. S., and O'Brien, M. J. (1973) Eur. J.
Biochem. 38, 201-211.
12. Rose, I- A., Warms, J. V. B., and Kosow, D. P. (1974)
Arch. Biochem. Biophvs. 164,729-735.
13. Holroyde, M. J., and Trayer, I. P. (1976) FEBS Lett. 62,
215-219.
14. Ureta, T. (1982) Como. Biochem. Physiol. 71B, 549-555.
15. Manning, T. A., and Wilson, J. E. (1984) Biochem.
Biophys. Res. Commun. 118, 90-96.
16. Anderson, C. M., Stenkamp, R. E., McDonald, R. C., and
Steitz, T. A. (1978) J. Mol. Biol. 123, 207-219.
17. Bennett, W. S., Jr., and Steitz, T. A. (1980) J. Mol.
Biol. 140, 211-230.
18. Griffin, L. D., Gelb, B. D.,Wheeler, D. A., Davison, D.,
Adams, V., and McCabe, E. R. B. (1991) Genomics 11,
1014-1024.
19. Redkar, V. D., and Kenkare, U. W. (1972) J. Biol. Chem.
247, 7576-7584.
20. Redkar, V. D., and Kenkare, U. W. (19751 Biochemistry
14, 4704-4712.
21. Subbarao, B., and Kenkare, U. W. (1977) Arch. Biochem.
Biophys. 181, 8-18.
22. Subbarao, B., and Kenkare, U. W. (1977) Arch. Biochem.
Biophys• 181, 19-25.
23. Swarup, G., and Kenkare, U. W. (1980) Biochemistry 19,
4048-4064.
24. Schirch, D. M., and Wilson, J. E. (1987) Arch. Biochem.
Biophys• 254, 385-396.
25. Schirch, D. M., and Wilson, J. E. (1987) Arch. Biochem.
Biophvs. 257, 1-12.
26. Dong, Q., Soans, C., Liu, F. and Fromm, H.J. (1990)
Arch. Biochem. Biophys. 276, 77-84.
27. White, T. K., and Wilson, J. E. (1989) Arch Biochem.
Biophvs. 274, 375-393.
28. Magnani, M., Bianchi, M., Casabianca, A., Stocchi, V.,
Daniele, A., Altruda, F., Ferrone, M. and Silengo, L.
(1992) Biochem. J. 285, 193-199.
29. Bork, P., Sander, C. and Valencia, A. (1992) Proc. Natl.
Acad. Sci. USA 89, 7290-7294.
30. Kabsch, W., Mannherz, H. G., Suck, D., Pai, E. F. and
Holmes, K. C. (1990) Nature 347, 37-44.
31. Flaherty, K. M., McKay, D.B., Kabsch, W. and Holmes, K.
C.r1991) Proc. Natl. Acad. Sci. USA 88. 5041-5045.
32. Liu, F., Dong, Q., Myers, A. M. and Fromm, H.J. (1991)
Biochem. Biophvs. Res. Commun. 177, 305-311.
33. Sanger, F., Nicklen, S., and Coulson, A. R. (1977) Proc.
Natl. Acad. Sci. USA, 74, 5463-5467.
34. Ferrari, R. A., Mandelstam, P. and Crane, R. K. (1959)
Arch. Biochem. Biophvs. 80, 372-377.
35. Bradford, M. M. (1976) Anal. Biochem. 72, 248-254.
36. Fromm, H.J. (1981) in The Regulation of Carbohydrate
Formation and Utilization in Mammals (C. M. Veneziale
Ed.),
University Park Press, Baltimore, pp45-68.
37. Ureta, T. (1975) in Isozymes III (C.L. Markert Ed.),
Academic Press, New York, pp 575-601.
38. Purich, D. L. and Fromm, H. J.(1971) J. Biol. Chem. 246,
3456-3463.
39. Ellison, W. R., Lueck, J. D. and Fromm, H. J. (1974)
Biochem. Biophys. Res. Commun. 57, 1214-1220.
40. Fromm, H. J. and Zewe, V. (1962) J. Biol. Chem. 237,
1661-1667.
41. Crane, R.K., and Sols, A. (1954) J. Biol. Chem. 210,
597-606.
42. Arora, K. K., Filburn, C. R. and Pedersen, P. L. (1993)
J. Biol. Chem. 268, 18259-18266.
43. Gerber, G., Preissler, H., Heinrich, R. and Rapoport, S.
M. (1974) Eur. J. Biochem. 45,39-52.
44. Lazo,P.A,, Sols, A. and Wilson, J. W. (1980) J. Biol.
Chem. 255, 7548-7551.
45. Wilson, J. E. (1984) in Regulation of Carbohydrate
Metabolism (Beitner, R., ed)
Vol. 1, pp. 45-85, CRC
Press, Inc., Boca Raton, FL.
46. Mehta, A., Jarori, G. K., and Kenkare, U. W. (1988) J.
Biol. Chem. 263. 15492-15497.
47. Jarori, G. K., Iyer, S. B., Kasturi, S. R., and Kenkare,
U. W. (1990) Eur. J. Biochem. 188, 9-14.
48. Sols, A. (1976) in Reflections on biochemistry
(Romberg, A., Horecker, B. L., Cornudella, L. and Oro,
J., eds), pp. 199-206, Pergamon, New York.
49. Solheim, L. P. and Fromm, H. J. (1981) Arch. Biochem.
Biophvs. 211, 92-99.
50. Schulz, G. E. (1992) Curr. Qpin. Struct. Biol. 2, 61-67.
51. Harrison, R. (1985) Crystalloaraphic Refinement of Two
Isozymes of Yeast Hexokinase and Relationship of
Structure to Function. Ph. D. Thesis, Yale University,
New Haven, CT.
34
CHAPTER 3. THE ATP-BINDING SITE OF HUMAN BRAIN HEXOKINASE AS
STUDIED BY MOLECULAR MODELING AND SITE-DIRECTED MUTAGENESIS^
A paper submitted to Biochemistry
Chenbo Zeng^'^ Alexander E. Aleshin^, Robert W. Harrison*,
Guanjun Chen and Herbert J.Fromm^'®
Abstract
The ATP binding site within the active site of brain
hexokinase is unknown since the crystal structure of the
hexokinase-ATP complex is unavailable.
It was found that the
ATP binding site of brain hexokinase is homologous to that of
actin, heat shock protein hsc70, and glycerol kinase.
Based
upon these similarities, the catalytic domain of human brain
hexokinase was modeled and the putative ATP-binding site was
proposed.
Site-directed mutagenesis was performed in the
putative ATP-binding site to test the proposed model.
^This research was supported in part by National Institutes of
Health Grant NS 10546, a grant from the United State Public Health
Service, and National Science Foundation Grant MCB-9218763.
^Graduate
student.
Research
associate
and
Professor,
respectively. Department of Biochemistry and Biophysics, Iowa State
University.
^Primary researcher and author
^Department of Pharmacology, Jefferson Cancer Institute,
Thomas Jefferson University, Philadelphia, PA 19107-5541
®To whom correspondence should be addressed
35
Asp532, which is thought to be involved in the interaction
with the
and Glu.
ion of the MgATP^" complex, was mutated to Lys
The k^at values decreased 1000-fold and 200-fold,
respectively for the two mutants.
Another residue, Thr680
was proposed to interact with the y-phosphoryl group of ATP
through hydrogen bonds, and was mutated to Val and Ser.
The
kcat value of the Thr680Val mutant decreased 2000-fold whereas
the kcat value of the Thr680Ser decreased only 2.5-fold,
implying the importance of the hydroxyl group.
The K„ values
for either ATP or glucose of all the above mutants showed
little or no change relative to the wild-type enzyme.
The
values for the glucose-6-P analogue, l,5-anhydroglucitol-6-P,
were the same as that of the wild-type enzyme, and the
inhibition was reversed by inorganic phosphate (Pi) for all
four mutants.
The circular dichroism spectra of the mutants
were the same as that of the wild-type enzyme.
The results
from the site-directed mutagenesis studies serve to support
the molecular model proposed for the ATP-binding site in
human brain hexokinase.
Introduction
Hexokinase I, brain hexokinase, is one of four
hexokinase isozymes found in mammalian tissue (Katzen &
Schimke, 1965).
It is believed to be the "pacemaker" of
glycolysis in brain tissue and the red blood cell (Lowry &
Passonneau, 1964; Rapoport, 1968).
The kinetic mechanism of
36
brain hexokinase is thought to be random bi bi (Ning et al.,
1969; Bachelard et al., 1971; Gerber et al., 1974), and a
mechanism for its molecular regulation has been proposed
(Ellison et al., 1975; Fromra, 1981; Wilson, 1984).
Brain
hexokinase exhibits regulatory properties that are unique
among the isozymes of hexokinase, i.e., it is potently
inhibited by its reaction product glucose-6-P, and this
inhibition can be reversed by physiological levels of
inorganic orthophosphate (Pi) (Rudolph & Fromm, 1971; Fromm,
1981; Ureta, 1975). Brain hexokinase consists of two
structural domains homologous to each other with about 50%
identity and to yeast hexokinase with about 30% identity
(Nishi et al., 1988). The catalytic properties of brain
hexokinase are associated with the C-terminal domain, while
the N-terminal domain is thought to be involved in molecular
regulation
(Nishi et al., 1988).
The structures of different crystal forms of the o: and
6 isozymes of yeast hexokinase were investigated by Steitz
and his collaborators (Anderson et al.,1978; Bennett et al.,
1980; Harrison, 1985).
glucose causes
They discovered that binding of
significant conformational changes in
hexokinase which narrows ("closes") the active site cleft of
the enzyme (Bennett et al., 1978). The published structures
of hexokinase b complexed with ortho-toluoylglucoseamine, and
hexokinase a complexed with glucose represent open and closed
conformations of the enzyme (Anderson et al.,1978; Bennett et
37
al., 1980; Harrison, 1985). In 1994 the computer generated
model for a mammalian hexokinase isozyme, glucokinase,
complexed with glucose was also published (Charles et al.
1994). In spite of success in understanding the interaction
of hexokinase with glucose, the interaction of the enzyme
with ATP is still unclear. Steitz and collaborators have
attempted to model ATP in the active site of the enzyme,
based on the crystal structure of open hexokinase b soaked
with 8-bromoadenosine monophosphate (Shoham & Steitz, 1980).
The authors suggested that ATP may be bound nonspecifically
to the open hexokinase form because the binding site for the
phosphate part of the ligand seemed to form during the
conformational transformation of the enzyme from the open to
the closed state (Shoham & Steitz, 1980).
Our study
represents the first attempt to model ATP within the active
site of mammalian hexokinase in its closed form.
Recently, it was reported that yeast hexokinase,
glycerol kinase, actin and heat shock protein hsc70 have
similar ATP-binding domain structures although there are no
overall primary sequence similarities among them (Bork et
al., 1992; Kabsch et al., 1990; Flaherty et al.,1991; Kabsch
& Holmes, 1995). This discovery suggests an alternative
approach for modeling ATP in the active site of hexokinase,
namely, by superimposing the homologous ATP-containing
structures on hexokinase. The model obtained using this
approach for the interaction of the C-terminal domain of
38
brain hexokinase with ATP pinpointed the putative ATP-binding
residues.
This paper presents kinetic studies on some
mutants of the presumed ATP-binding residues.
Our results
provide strong support for our ATP-binding site model of
human brain hexokinase.
Experimental Procedures
Materials—Affi-Gel Blue Gel and Bio-gel hydroxyapatite
were obtained from Bio-Rad Laboratories.
Transformer™ Site-
Directed Mutagenesis Kit (2nd version) was a product of
Clontech.
The Magic Minipreps DNA purification system was a
product of Promega.
Oligonucleotide synthesis and nucleotide
sequence analysis were done by the Iowa State University
Nucleic Acid Facility.
Xho I was from Promega.
from Novagen.
Nru I was from New England Biolabs.
The pET-lla plasmid was obtained
Escherichia coli strains BL21(DE3) and TG-1
cells were obtained from Amersham Corp.
Ampicillin,
phenylmethylsulfonyl fluoride (PMSF)®, ATP, NADP, NADPH,
glucose-1,6-bisphosphate (Glu-l,6-BP), deoxyribonuclease I
(DNase I), Isopropyl-l-thio-6-D-galactopyranoside (IPTG), and
1,5-anhydro-D-sorbitol were products of Sigma.
Modeling the brain hexokinase complexed with ATP and
jIg2+_The alignment of the sequences was made by analogy with
®The abbreviations used are: PMSF, phenylmethylsulfonylfluoride; IPTG, Isopropyl-B-D-thiogalactopyranoside; Glu1,6-BP, Glu-1,6-bisphosphate; Glucose-6-P,glucose-6-phosphate; 1,5an-G6P, 1,5-anhydroglucitol-6-phosphate.
39
the structure-based alignment of human glucokinase and yeast
hexokinase B reported earlier (Charles et al., 1994).
The
program AMMP was used to build the models (Harrison, 1993).
The force field was the sp3 variation of the UFP potential
set (Harrison, 1993; Rappe et al., 1992).
The amortized fast
multipole algorithm was used for the non-bonded and
electrostatic terms and no cut off radius was applied.
Crystallographically ordered solvent from yeast hexokinase
was included.
The open form was modelled by fixing the
conserved atoms at their observed positions and building the
remainder of the structure.
minimized.
Then the whole structure was
A homotopy method (Harrison, 1995) was used to
minimize the structures.
In the homotopy method the targets
for the minimization are gradually shifted from the current
erroneous positions to the desired targets.
The steps are
controlled by a critical path parameter A. which is slowly
varied from 1 to 0.
The homotopy was applied to the bonds
and angle terms with the pyramid height torsion and nonbonded terms always being minimized.
The path was tracked by
minimization which approximates tracking a zero in the first
derivative of the energy.
was used.
Conjugate gradient minimization
The structure was minimized in two passes where
the tracking parameter
steps of 0.05.
X was varied form 0.95 to 0 in 20
Procheck (Laskowshi et al., 1993) was used to
check the stereochemistry at the end of the process and no
significant errors were found.
40
The closed form of hexokinase was initially modelled in
the same manner, but starting from the yeast hexokinase A
coordinate set.
There is a region in one of the helices in
the small domain of yeast hexokinase A which was poorly
determined in the electron density (Harrison, 1985).
When
the model in that region was examined, it was found to be
unsatisfactory, e.g., hydrophilic amino acids such as
arginine were pointing into the domain and hydrophobic acids
such as tyrosine or phenylalanine were pointing out into the
solvent.
The stereochemistry in that region was also poor.
Therefore, it was decided to model the closed structure by
using the yeast coordinates as guides for the backbone and
deforming the open structure into the closed form.
The same
homotopy method was used with the homotopy applied to the a
carbon restraints as well as the bonds and angles.
The two
modelled structures were similar in the small domain except
where the poorly defined region was, and in that region the
new model was much more consistent with the expected
distribution of amino acid side chain positions.
The initial positions and conformations of MgATP^"
complexes in the active site of brain hexokinase were
obtained by superimposing the conserved residues of the ATP
binding site
of the enzyme
(Bork et al., 1992)) and the
corresponding residues of either actin, heat shock protein,
or glycerol kinase complexed with
ligands (PDB entries are
lATN, INGJ, IGLC, respectively). The model of the complexes
41
of the closed form of brain hexokinase and
ATP, which were
built either from the ADP complexes of glycerol kinase or
from the ATP complexes of actin, were further energy
minimized
with the same AMMP program.
Expression of the Wild Type and Mutant Human Brain
Hexokinase I—The cDNA for human brain hexokinase I has been
cloned into the expression vector pET-lla to generate pETlla-HKI in our laboratory (Liu et al., 1991).
pET-lla-HKI
was transformed into E. Coli strain BL21(DE3), a phage klysogen containing the T7 RNA polymerase gene under control
of the lacUVS promoter.
A 5-inL pregrowth culture of the
transformed strain in M9 medium plus 40 mg/L ampicillin, 1 mM
MgS04 and 4mg/L glucose was grown overnight and then added to
500 ml of the same medium.
The culture was shaken at 37° C
to early log phase (A6oo=0.4), and IPTG was then added to a
final concentration of 0.4 mM to induce the T7 RNA polymerase
gene.
Cultures were shaken vigorously for an additional 24
hours at 22° C.
Purification of the Wild Type and Mutant Forms of Human
Brain Hexokinase I—The cells containing the overexpressed
hexokinase were collected by centrifugation.
About 30-gram
wet cells (~30 mL) were suspended in a 30-mL of suspension
solution containing 50 mM Tris-HCl (pH8.0)-10 mM glucose-0.5
jmM EDTA-lmM mercaptoethanol-2 mM PMSF.
The cells were broken
by French Press at a pressure of lOOOpsi.
repeated twice.
This procedure was
0.9-mL of DNasel solution (0.15-mL 10 mg/mL
42
DNasel stock plus 0.65-inL 1 M MgCl2) was mixed with the
broken cells for 30 minutes.
The broken cell suspension was
subjected to centrifugation for 50 minutes at 12,000 rpm.
The supernatant fluid was dialyzed against 2 L of the
dialysis buffer containing 50 mM Tris-HCl (pH7.0)-10 mM
glucose-0.5 mM EDTA-1 mM mercaptoethanol. The buffer was
changed twice.
The dialyzed cell extract was loaded onto a
X 13 cm Affi-Gel Blue Gel column (Wilson,1989) and washed
with the same solution that was used for dialysis.
The
column was further washed with the buffer having the above
composition tut at pH 8.0, and then with the bufter
containing 50 mM Tris-HCl (pH 9.0)-10 mM glucose-0.5 mM EDTA
1 mM mercaptoethanol-20% (v/v) glycerol.
Human brain
hexokinase I was eluted with 1.5 mM glucose-6-P in the same
pH 9.0 buffer.
The fractions containing human brain
hexokinase were identified by SDS-polyacrylamide gel
electrophoresis, pooled
and subjected to ultrafiltration.
5-10 ml of concentrated enzyme solution was loaded onto a 1
30 cm Bio-gel hydroxyapatite column equilibrated with buffer
containing 25 mM KPi-1 mM mercaptoethanol.
eluted with
The protein was
180 mL of a linear gradient solution of low
ionic strength buffer containing 25 mM KPi-1 mM
mercaptoethanol and 180 mL of high ionic strength buffer
containing 0.3 M KPi-1 mM mercaptoethanol.
The fractions
containing pure human brain hexokinase were identified by
SDS-polyacrylamide electrophoresis, collected, concentrated.
43
and stored in 50% (v/v) glycerol at -20" C.
Site-directed Mutagenesis—Mutagenesis was performed by
following the instructions provided with Tansformer™ SiteDirected Mutagenesis Kit (2nd version) from Clontech.
The
oligonucleotides used for mutagenesis are as follows:
5'-
TTCTTGGCCCTGAAGCTTGGAGGAAC-3' for Asp532Lys, 5'GCCCTGGAACTTGGAGG-3' for Asp532Glu, 5'-TTGTTGGGGTCGGCAGCAATG3' for Thr680Val, and 5'-TTGTTGGGTCCGGCAGCAATG-3' for
Thr680Ser, where the underlines represent the mutated bases.
The selective oligonucleotide sequence was chosen at the Nrul
restriction site in the vector pET-lla and designed to change
the Nrul site to a Xhol site.
The sequence of the selective
oligonucleotide is 5^-CAGCCTCGCCTCGAGAACGCCAG-3^. where the
underline represents the Xhol site.
The DNA sequence of the
mutated gene was verified by DNA sequencing using the method
of fluorescent dideoxychain termination sequencing.
Hexokinase Assay—Hexokinase activity was determined
spectrophotometrically as described previously (Liu et al.,
1991).
For kinetic characterization, the amount of the wild-
type and mutant human brain hexokinase was adjusted to give
an activity of approximately 5 x lO"^ AODs^o/roin.
Human brain
hexokinase enzyme was dialyzed against a solution containing
50 mM Tris-HCl (pH 7.0)-lmM mercaptoethanol to remove Pi.
Glucose-6-phosphate dehydrogenase was dialyzed against the
same solution to remove (NH4)2S04.
Since glucose-6-phosphate
dehydrogenase can use glucose as its substrate at high levels
44
of the sugar, the amount of glucose-6-phosphate dehydrogenase
used was as small as possible in order to preclude glucose
oxidation.
0.1 unit of glucose-6-phosphate dehydrogenase was
used in 1-ml of the assay system.
Preparation of 1.5-an-G6P—1.5-an-G6P was prepared as
described previously (Ferrari et al., 1959).
Determination of Protein Concentration—Protein
concentration was determined by the method of Bradford
(Bradford, 1976), using bovine serum albumin as a standard.
Circular Dichroism Measurement—Circular Dichroism (CD)
spectra were recorded using a Jasco J710 CD spectrometer as
described previously (Zeng & Fromm, 1995).
Results and Discussion
Purification of wild-type and Mutant Human Brain
Hexokinase—The purification procedure outlined in this report
is greatly simplified relative to previously reported methods
(Zeng & Fromm, 1995) and the enzyme yield is increased (2030%).
Approximately, 20 mg of pure enzyme can be obtained
from 10 liters of E. coli culture.
The procedure requires
only two columns (Affi-Gel Blue Gel and Bio-Gel
hydroxyapatite), dialysis and ultrafiltration.
Fig. 3.1
shows the SDS-PAGE at each step of purification.
The mutant
enzymes, Asp532Lys, Asp532Glu, Thr680Val, and Thr680Ser, bind
to the Affi-Gel Blue-Gel column, implying that their ATP
binding ability is retained after mutagenesis.
45
Modeled interaction of brain hexokinase with ATP and
Mg2i—The sequence alignment of the catalytic domain of human
hexokinase with yeast hexokinase isomers A and B revealed 31%
and 33% identity, respectively. The homology is especially
high in the elements of secondary structure and in the
regions
of interactions with substrates (Fig. 3.2). The
model of the closed form of brain hexokinase, which was built
from its open form by bending to yeast hexokinase A, is as
similar to the prototype structure as the
model built
directly from hexokinase A; however, the geometry of its
small lobe (Fig. 3.3) is significantly improved.
Our modeling of the ternary complex of brain hexokinase
was based on the assumption that the position of ATP and Mg^*
relative to the conservative residues of
the active site
must be consistent in all homologous phosphotransferases. The
crystal structures of glycerol kinase ligated with ADP and
Mg^^ (Hurley et al., 1993), the ATP-binding domain of actin
ligated with ATP and Ca®"^ (Kabsch et al., 1990), and the ATPbinding domain of heat shock protein ligated with ANP (5'adenylyl-B,y-imidodiphosphate) and Mg^"^ (Flaherty et al.,
1994) (Protein Data Base entries are IGLC, lATN, INGJ,
respectively) were used for modeling. The superimposition of
these enzymes on the model of the closed form of the
catalytic domain of brain hexokinase demonstrated that all
three ATP and Mg^"^ analogues have similar positions in the
active site of brain hexokinase (Fig. 3.4).
ADP from
46
1 2 3 4 5
a b -
c -
It
d e -
Fig. 3.1. SDS PAGE patterns for each step of the
purification of wild-type human brain hexokinase. Lane 2,
cell-free extract, lane 3, the enzyme fraction after the
Affi-Gel Blue Gel column, and lane 4, the enzyme fraction
after the Bio-Gel hydroxyapatite column. Lanes 1 and 5,
protein standards; the molecular masses for the bands a,
c, d, and e are 200, 116, 97.4, 66.2, and 45 kDa,
respectively.
47
470
_
480 „ _
460
VQY R L A IQ H RBOfclE TLA HQhPT K^MQL EC
M!30 V P K II MPglgHQL E DMiSt V OS ETMR kC
V P K1 II MQBffal P E K 1 QtlIuP TliiTSQAC
BRAIN HK
yeastihka
YEAST.HKB
30
)
40
ai
pi
a2
520
IR RTQDSTlEKJGDF L AaDLGGTN
ME F V P T G E 4G[3iai: AtlD L G GT N
IMD FQTQkIE B G D F L A l l D L G G T N
, ^ „ ^0.
„
570
V^MH^K I QArTai EH MQGTG
F P T T Q S K QK L E3H OMR T T K H Q
FIqJt TIfljs kHrLho aImIr T T Q N P
110.
P3
P2
120
P4
GGG
590
irrav MG
I K GP R
ImEEQE L L NT K DT
jaEQFPQGI SEP_
140
150
WT KGF K A T D C V
HBiiliilBi
IORQR WWTT KK GG FF ID01ISOI PP NN VI ENE_
P6
P5
660
IE F D
pr
^0
200
P P P GGGG
iGT G
naM^En
^Qy
p V cs 0 1
GT G
IGT G
^
MyIdIvIcIs oil
240
230
,
__700
QQ
GQ
Ik l aIdd 1 p s N s p
3k l sQd I P P S A P
260
pio
L N A KmR
QEIA
m
i
RDA
QK E
QKQ
P7
a5
•I
510
IQT H N NKlVMKlilLtJ:
3KG
GNiPmiQi
3
kg
IgInm Ipmi n
[3Nt
aL V QTOIF
IL E L THaaaoa
N ECtHLlMfL
SIla •MraMYCiDBiaf a
PRP
270
Pll
—
78(
'GQl SETMHTRGI
IK OQD L S K LtaOP Y
1N0D LSKFDKPF_
330
340
RAAAAAA
KF LSQ
SIY P A R
SMPAR
800
iSiaR
lALLQVRA) LOQL
DIJP FENLEOTDDI FQKOF
EiaP FENLEDTDOLFQNEF
V kt
0
op S)
L P E RK
VGI EIRK
all
850
_
11 R E N R G L D R L ^
YKfr
IRG
IRG
YKEL
410
KL H
NKY
NRY
SRI MHQTwSiLSP
iaCNmSFl&
KEAAAKGLRDI'YGWTGENA SISD P I R I V
K 6 K A A N A L R I D t Y G V<T O T S L D D Y P U k I k J
al2
LS
PA
p a|
Pl3
900
,0
KHEHniTQV]iGVfR L RT |4A]
ASSivtlAraLf S EK R 1 A SQ]
aKCTIVMAEil |A QKIRI aIlaGl
4"'0
Fig. 3.2. Structural alignment of amino acid sequences
of the catalytic domain of human brain hexokinase and yeast
hexokinases A and B. The residues mutually identical with
those of brain hexokinase are outlined in black, and
homologous residues are in shaded boxes. The elements of
secondary structure are from the crystal structure of yeast
hexokinase B. G, P, R, and A on the top of the brain
hexokinase sequence indicate residues forming the binding
sites for glucose, phosphates, ribose, and adenine,
respectively. The picture was prepared using
ALSCRIPT(Barton, 1993).
48
Fig. 3.3. Stereoview of the Co backbone of the modelled
structure of the catalytic domain of brain hexokinase. The
domain has a closed form. ATP, glucose, and Mg^"^ ion bound
in the cleft are shown with bold lines. The conservative
residues within phosphotransferases are shown with filled
circles.
49
glycerol kinase is shifted
approximately 2.5
A
away from the
ATP molecules of actin and heat shock protein structures
which have almost identical positions.
The difference in the
positions is greater for the adenine rings of the ligands
(rms deviation is 3.0
A)
than for B-phosphates and Mg^^ (the
deviations are less than 2
A).
This result correlates with
the significant discrepancy between the adenine binding sites
of glycerol kinase and actin. Adenine of ADP from glycerol
kinase fits better into the active site of brain hexokinase.
This agrees with the similarity of the adenine binding sites
of these enzymes. Mg^* and the phosphates of all 3
superimposed ATP analogues also satisfactorily fit the model.
However, all superimposed ATP analogues have Thr863
overlapping with ribose.
In all phosphortransferases except
hexokinases, this position in the sequences is represented by
a conserved glycine (Bork et al., 1992). The bad fit of
ribose suggests that either brain hexokinase changes the
orientation of Thr863 upon the binding of ATP, or that ATP in
hexokinases has a slightly different conformation than in the
other enzymes.
The energy minimization of ATP in the active site of
brain hexokinase removed the bad contact without a
significant change in the position of the ligand. We used ATP
and
ligands.
from superimposed actin as a starting model for the
After the energy minimization, adenine of ATP
shifted to a position close to that of adenine from the
50
Fig. 3.4. Stereoview of the ATP binding site of brain
hexokinase(closed form) with ATP analogues superimposed from
different crystal structures. The backbone of the enzyme is
represented by smoothed coils. The side chains of
conservative residues are drawn as ball-and-sticks. Glucose
is shown in black. (A) labels adenine of 8-bromoadenosine
monophosphate from the crystal structure of yeast hexokinase
B(Shoham & Steitz, 1980). (B) labels ADP and Mg^^Cfilled
circle) from glycerol kinase(Hurley et al., 1993). (C)
labels ATP and Ca^^(open circle)from actin(Kabsch et al.,
1990). The figure was prepared with MOLSCRIPT(Kraulis,
1991).
51
glycerol kinase structure. However, the yphosphate resumed
its initial position. The test minimization of the complex,
with ADP and
from glycerol kinase as a starting model
gave similar results.
The energy minimized model of the active site of the
enzyme with bound glucose, ATP and Mg^^ is presented in Fig.
3.5. The ATP molecule is positioned in the cleft between two
lobes of the enzyme. Adenine is bound to the big lobe and
occupies a cavity between helix a9 (residues 784-788) and the
loop, connecting strand 612 and helix al2 (residues 863-869).
Ribose is bound between side chains of Glu 783 and Thr 863.
The loop residues 863-869 have a similar function in all
homologous phosphotransferases, i.e.. they form the basis of
the adenosine-binding site.
According to our modeling, one
face of adenine is exposed to the solvent. This is quite
different from other structurally known phosphotransferases.
It is possible that upon the binding of ATP, hexokinase
undergoes further conformational change, leading it to a more
closed conformation (Kabsch & Holmes, 1995).
The phosphate binding site consists of residues from two
lobes or the catalytic domain of brain hexokinase.
At one
side, phosphates interact with residues 680 and 681 of the Bturn between strands 6-9 and 6-10, and at the other side,
they interact with the loop connecting strands 6-2 and 6-3.
Both elements are conserved within all homologous
phosphotransferases (Bork et al., 1992; Kabsch et al., 1990;
52
Fig. 3.5. Stereodrawing of the ATP-binding site of the
model of the ternary complex of brain hexokinase. Glucose
and ATP are drawn in ball-and-stick, the filled balls are N,
O, or Mg^"^ atoms. The protein part of the complex is drawn
with thin lines(except side chains of Asp532 and Asp657).
The hydrogen bonds and salt linkages between the protein and
ligands are shown as dash lines. The figure was prepared
with MOLSCRIPT(Kraulis, 1991).
53
Flaherty et al., 1991). The
e -amino
group of Lys621 is 4.6
A
from the y-phosphate. The modeling suggests that this residue
can be involved in ATP binding.
In our model Mg^" is coordinated by OIB of the
B-phosphate, and 03G of the y-phosphate of ATP and it is
positioned relatively close to Asp532. The model suggests
that Asp532 coordinates Vlg^* either by direct ionic
interactions or through a mediating water molecule.
This
finding supports the hypothesis (Wilbands et al., 1994;
Flaherty et al., 1994) that the catalytic role of this
aspartate in phosphotransferases is to stabilize the correct
position of y-phosphate through its interaction with Mg^^.
Our modeling of ATP bound to the closed form of
hexokinase supports the suggestion of Steitz and
collaborations that the correct binding of the
tripolyphosphate part of ATP requires the closed form of
hexokinase (Shoham & Steitz, 1980).
The positions of all
three superimposed ATP analogues, being in a good agreement
with each other, noticeably differ from that of
8-bromoadenosine monophosphate in the crystal structure of
open yeast hexokinase
(Fig. 3.4).
The y-phosphate of ATP in
the crystallographically determined complex corresponds to
a-phosphate in our model and is about 6
A
away from the 06
hydroxyl of glucose. A y-phosphate of ATP in our model has a
more realistic distance from 06 of glucose equal to 3.9
A.
Assuming that the two models represent the interaction of ATP
54
with open and closed forms of hexokinase, respectively,
we
can conclude that interactions between the phosphate part of
ATP and hexokinase are responsible for the formation of the
productive ATP-hexokinase complex, and that the phosphate
binding site of hexokinase is not formed correctly
in the
open form of the enzyme.
The model also explains our previous finding (Zeng &
Fromm, 1995) that shows the importance of Arg539 for
catalysis.
Arg539 interacts with a- and 6-phosphate groups
of ATP and may stabilize the transition state.
In order to
test the predictions of the model, another two residues
within different loci of the ATP binding site were mutated.
Site-directed Mutagenesis of
ASP532
in the Putative ATP-
bindina Site—Based on our model of the ATP binding site of
human brain hexokinase, Asp532 might be involved in binding
of the Mg^^ ion in the MgATP^" complex.
To test this
hypothesis, Asp532 was mutated to Lys and Glu.
The kinetic
parameters for the mutants, Asp532Lys and Asp532Glu, and the
wild-type enzyme, were determined and are shown in Table 3.1.
The mutants, Asp532Lys and Asp532Glu, showed only small
changes in K, values for glucose and ATP, compared with the
wild-type enzyme.
However, the k^^t values of the mutants,
Asp532Lys and Asp532Glu, exhibited 1000-fold and 200-fold
decreases relative to the wild-type enzyme, respectively. The
mutant Asp532Lys was inhibited by the glucose-6-P analogue
l,5-an-G6P, with the same Kj values as that of the wild-type
Table 3.1.
A Comparison of the kinetic data of the wild-type and the mutant forms of
hexokinase
K„(G1U)
K„(ATP)
(/IM)
(mM)
K5(l,5-An-G6P)
Hexokinase
(MM)
(s-1)
26.9
wild-type
56 + 5.6
1.0 ± 0.2
26 ± 4.8
Gly534Ala
287 + 22.2
2.2 + 0.13
82 + 16.5
Gly679Ala
42 ± 6.4
0.23 + 0.39
27 + 6.0
19.3
Arg539Lys
52 + 0.39
0.43 + 0.11
19 + 4.4
2.2
Tyr686His
43 ± 4.1
0.35 ± 0.06
9 + 0.8
35.5
0.0063
The values shown are the mean ± SME of at least two separate determinations.
kinetics analysis was done with the computer program "Kinetics".
The
56
enzyme.
Pi reversed the inhibition.
The dramatic decreases
in activities of hexokinase mutants Asp532Lys and Asp532Glu,
illustrates the importance of Asp532 for the catalytic
reaction of human brain hexokinase. Based on our model we can
suggest two ways that Asp532 may be involved in catalysis:
(1) its negative charge may be necessary for the formation of
the correct electrostatic environment in the active site, (2)
the interaction between Asp532 and the Mg^^ ion may stabilize
the y-phosphate of ATP in the vicinity of 06-hydroxyl of
glucose.
The recent report of Baijal and Wilson (Baijal &
Wilson, 1995) that Asp532Asn has significantly reduced
activity indicates that, indeed, the charge on this group is
important for catalysis.
Our results, support the importance
of the negative charge at position 532, but also demonstrate
that the size of the charged group is important as well.
In heat shock protein, hsc70, the residue corresponding
to Asp532 of human brain hexokinase is Aspll, whereas in
actin it is AsplO.
Flaherty et al. (Wilbands et al., 1994;
Flaherty et al., 1994) mutated AsplO to Asn and Ser.
The k^^^
values decreased to about 1% of the wild-type enzyme, whereas
the K„ value for ATP increased more than ten-fold.
In the
crystal structure of hsc70, the carboxyl group of AsplO
ligates to the hydrated Mg^*.
The kinetic data of mutants
AsplOAsn and AsplOSer as well as other mutants suggest that
the ability of the Mg^^ ion to position the Y~Phosphate for
nucleophilic attack is
critical.
The conclusions of these
57
investigators is similar to that which we arrived at for the
function of Asp532 in human brain hexokinase.
The K„ values for ATP for the mutants Asp532LYs and
Asp532Glu hexokinase are the same as for the wild-type
enzyme, indicating the interaction between Asp532 and the
Mg^"^ ion of the MgATP^' complex is not responsible for ATP
binding affinity.
Our previous results (Rudolph & Fromm,
1971; Kosow & Ross, 1970) indicate the Kj value for the ATP""
is about the same as the K„ value for MgATP^", a finding
consistent with our present result.
The fact that the kcat values decreased signi^icaBtly for
mutants Asp532Lys, and Asp532Glu relative to wild-type
hexokinase, whereas the K„ values remain relatively constant,
suggests that the major effect of these mutations occurs in
the conversion of the enzyme-Mg^'^ATP-glucose complex to the
analogous product ternary complex, i.e., in the transition
state.
This conclusion is alluded to based on the kinetic
mechanism for brain hexokinase which is rapid-equilibrium
random bi bi (Ning et al., 1969).
In this kinetic mechanism
the breakdown of the enzyme-Mg^^ATP-glucose complex to
enzyme-Mg^*-ADP-glucose-6-P is rate limiting, and k^at is a
reflection of the nature of the transition state, but not its
structure.
Site-directed Mutagenesis of Thr680—Our model of the
ATP-binding site of human brain hexokinase predicted that the
hydroxyl group of Thr680 may form hydrogen bonds to y-
58
phosphoryl group of ATP (Fig. 3.6) and thus may be important
for the catalytic reaction of hexokinase.
was mutated to Ser and Val.
Therefore, Thr680
Although the K„ values of both
mutants for either glucose or ATP showed only slight or no
change, the
values for Thr680Val and Thr680Ser mutants
exhibited 2000-fold and 2.5-fold decreases, respectively
(Table 3.1).
Since CD spectra for Thr680Val and Thr680Ser
showed no difference relative to that of the wild-type
enzyme, we suggest that no secondary conformation change
occurs after mutagenesis.
The dramatic decrease in the ke«t
value with no change in the K„ values for glucose and ATP
suggests that the Thr680 is important for catalysis rather
for binding of ATP.
Thr680 and
If the proposed interaction between
OIG of y-phosphate is correct, then this
interaction may be involved in stabilization of the
transition state of the reaction.
Moreover, the function of
Thr680 may be similar to that proposed for Asp532, namely, it
may participate in the stabilization of the vpliosphate in
the vicinity of 06-hydroxyl group of glucose.
It was found (Pilkis et al., 1994) that mutations in the
human glucokinase gene are a common cause of an autosomal
dominant form of non-insulin-dependent (type 2) diabetes
mellitus.
One mutant, Thr228Met, causes about a 2500-fold
decrease in the k<,at value but no change in the K„ values
either for glucose or for ATP relative to the wild-type
glucokinase.
Thr228 corresponds to Thr680 in human brain
59
Ar?539
Ttii-863
-?
HP
y
O
H
•'
HW.
Asp532
H
o
V
H^
Hgl
!
O 6 b O
\/ \ /
"itf
O
OH
y
Thr680
Aap657
V
'o
/
H
„mH
H
O
OH
HO
^OH
I<ys621
I G1
61u783
Fig. 3.6. Schematic drawing of interactions of brain
hexokinase with ATP, as predicted by modeling.
60
hexokinase.
The kinetic properties of Thr228Met mutant is
similar to those of the Thr680Val mutant in human brain
hexokinase.
Assuming human glucokinase has the same ATP-
binding site as that of human brain hexokinase, Thr228 may
interact with the ATP in the same way as Thr680 does.
The kinetic studies of mutants of residues Asp532,
Thr680, and Arg539 of human brain hexokinase, which according
to our modeling are involved in binding of tripolyphosphate
portion of ATP, indicate that these residues are important
for the catalysis.
These results provide strong support for
the method of modeling the ternary complex of hexokinase with
ATP and glucose presented in this report.
All studied
mutants have one common and intriguing feature, namely: in
spite of the involvement of these residues in binding of ATP,
their mutation significantly reduces only enzyme activity but
does not alter ATP affinity.
The modeling suggests that all
these residues may be involved in the stabilization of the
position of y-phosphate of ATP in the vicinity of 06 of
glucose.
Thus, the energy of binding of the phosphate part
of ATP is presumably used to promote the transition state,
while the energy of ATP binding comes from nonspecific
interactions such as the entropy changes during the
substitution of water molecules from the active site.
61
Acknowledgments
We thank Ms. Ju-Chi Cheng for her excellent technical
assistance.
References
Anderson, C. M., Stenkamp, R. E., McDonald, R. C., and
Steitz, T. A. (1978) J. Mol. Biol. 123, 207.
Bachelard, H. S., Clark, A. G., and Thompson, M.
F.(1971) Biochem. J. 123, 707.
Baijal, M., and Wilson, J. E. (1995) Arch Biochem.
BiopTlvs. 321, 413.
Barton, G. J. (1993) Protein Engineering. 6, 37.
Bennett, W. S., Jr., and Steitz, T. A. (1978) Proc. Natl.
Acad. Sci. USA 75, 4848.
Bennett, W. S., Jr., and Steitz, T. A. (1980) J. Mol.
Biol. 140, 211.
Bork, P., Sander, C. and Valencia, A. (1992) Proc. Natl.
Acad. Sci. USA 89, 7290.
Bradford, M. M. (1976) Anal. Biochem. 72, 248.
Charles, R. St., Harrison, R. W., Bell, G. I., Pilkis,
S. J., and Weber, I. T. (1994) Diabetes 43, 784.
Ellison, W. R., Lueck, J. D., and Fromm, H. J.(1975) J.
Biol. Chem. 250, 1864.
Ferrari, R. A., Mandelstam, P., and Crane, R. K. (1959)
Arch Biochem. Biophys. 80, 372.
62
Flaherty, K. M., McKay, D. B., Kabsch, W. and Holmes, K.
C.ri991') Proc. Natl. Acad. Sci. USA 88. 5041.
Flaherty, K. M., Wilbands, S. M., Flaherty, C. D., and
McKay, D. B. (1994) J. Biol. Chem. 269, 12899.
Fromm, H.J. (1981) in The Regulation of Carbohydrate
Formation and Utilization in Mammals (C. M. Veneziale
Ed.), University Park Press, Baltimore, pp45-68.
Gerber, G., Preissler H., Heinrich, R., and Rapoport, S.
M.(1974) Eur. J. Biochem. 45, 39.
Harrison, R. (1985) Crystalloaraphic Refinement of Two
Tsozymes of Yeast Hexokinase and Relationshi-p of
Structure to Function. Ph. D. Thesis, Yale University,
New Haven, CT.
Harrison, R. W. (1993) J. Comput• Chem. 14, 1112.
Harrison, R., W. (1995) The manuscript in preparation
Hurley, J. H., Faber, H. R., Worthylake, D., Meadow, N.
D., Roseman, 5., Pettigrew, D. W., and Remington, S. J.
(1993) Science 259, 673.
Kabsch, W., Mannherz, H. G., Suck, D., Pai, E. F. and
Holmes, K. C. (1990) Nature 347, 37.
Kabsch, W., and Holmes, K. C.(1995) FASEB 9, 167.
Katzen, H. M. and Schimke, R. T. (1965) Proc. Natl.
Acad. Sci. USA 54, 1218.
Kosow, D. P. and Ross, I. A. (1970) J. Biol. Chem. 245,
198.
Kraulis, P. J., (1991) J. of Applied Cryst. 24, 946.
63
Laskowshi, R. A., MacArthur, M. W., Moss, D. S., and
Thornton, T. M.(1993) J. Appl. Crvst. 26, in press
Liu, F., Dong, Q., Myers, A. M. and Fromm, H.J. (1991)
Biochem. Biophys. Res. Commun. 177, 305.
Lowry, O.H., and Passonneau, J.V. (1964) J. Biol. Chem.
239, 31.
Ning, J., Purich, D. L., and Fromm, H. J. (1969)
J.
Biol. Chem. 244, 3840.
Nishi, S., Seino, S., and Bell, G. I. (1988) Biochem.
Biophvs. Res. Commun. 157, 937.
Pilkis, S. J., T^eber, I. T., Harrison, R. W., and Bell,
I. (1994) J. Biol. Chem. 269, 21925.
Rapoport, S. (1968) Essays Biochem. 4, 69.
Rappe, A. K., Casewit, C. J., Colwell, K. S., Goddard
III, W. A., and Skiff, W. M. (1992) J. Am. Chem. Soc.
114, 10024.
Rudolph, F. B., and Fromm, H. J. (1971) J. Biol. Chem.
246, 6611.
Shoham, M. and Steitz, T. A. (1980) J. Mol. Biol. 140,
1-14.
Ureta, T. (1975) in Isozymes IIIfMarkert. C. L.k ed) pp.
575-601, Academic Press, Inc., New York
Wilbands, S. M., Flaherty, C. D., and McKay, D. B.
(1994) J. Biol. Chem. 269, 12893.
64
Wilson, J. E. (1984) in Regulation of
Metabo1ismfBeitner. R., ed)
Carbohydrate
Vol. 1, pp. 45-85, CRC
Press, Inc., Boca Raton, FL.
Wilson, J. E. (1989) Preparative Biochemistry 19(1), 13.
Zeng, C., and Fromm, H. J. (1995) J. Biol. Chem. 270,
10509.
65
CHAPTER 4. A STUDY OF THE ATP-BINDING SITE OF HUHAN BRAIN
HEXOKINASE^
A paper to be submitted to the Arch. Biochem. Biophysics.
Chenbo Zeng^, Guanjun Chen and Herbert J. Fromm^'^
Abstract
The ATP-binding site of human brain hexokinase has been
modelled and confirmed by site-directed mutagenesis.
Several
residues located in the proposed ATP-binding site are studied
in the present paper.
Four mutants were prepared: Gly534Ala,
Gly679Ala, Arg539Lys and Tyr686His.
The
values for the
Gly534Ala and Arg539Lys mutants decreased 4600-fold and 12fold, respectively,
whereas the k„t values for the Gly679Ala
and Tyr686His remained about the same as that of the wildtype enzyme.
The K„ values for either glucose or ATP did not
show dramatic changes for any of the four mutants relative to
the wild-type enzyme.
glucose-6-P with the
All the mutants can be inhibited by
values similar to the wild-type
'^This researach was supported in part by Research Grants
NS10546 from the National Institue of Health, United State Public
Health Service, and Grant MCB-9218763 from the National Science
Foundation.
^Graduate student and Professor, respectively. Department of
Biochemistry and Biophysics, Iowa State University.
Research
conducted and manuscript written by Zeng with assistance from Chen
and Fromm.
^To whom crrespondence should be addressed.
66
enzyme, and the inhibition can be released by inorganic
phosphate.
The circular dichroism and fluorescence spectra
did not show significant changes for the mutants compared
with the wild-type enzyme.
Heat stability experiment showed
that only Tyr686His is less thermostable than the wild-type
enzyme.
Based on our molecular model of the ATP-binding
site, the small side chain of Gly534 interacts with ATP and
is essential to make appropriate room for binding ATP. Arg539
binds to the a- and 6- phosphate groups of ATP.
We suggest
that the positive charge of Arg539 is important by
stabilizing the transition state of the enzyme.
The small
side chain of Gly679 may be important for the correct folding
of hexokinase.
Tyr686 appears to be responsible for
thermostability.
Introduction
Hexokinase I, brain hexokinase, is one of four
hexokinase isozymes found in mammalian tissue(l).
It
catalyzes the phosphorylation of glucose using ATP as a
phosphate donor.
It is the first rate-limiting reaction of
the glycolytic pathway in brain tissue and the red blood
cell(2,3).
Brain hexokinase exhibits a unique regulatory
properties, i.e., it is potently inhibited by its reaction
product glucose-6-P relative to the other hexokinase
isozymes, and this inhibition can be reversed by
physiological levels of inorganic orthophosphate(Pi)(4-6).
67
Hexokinase isoforms I, II, and III exist as a monomer with
molecular weights of approximately 100,000 and show
similarity in amino acid sequence(7).
Hexokinase isoform IV,
glucokinase, has a molecular weight of 50,000, similar to
that of yeast hexokinase.
Moreover, isozymes I, II and III
contain N-terminal and C-terminal domains in which the
domains exhibit similar amino acid sequences to each other
and to glucokinase and yeast hexokinase, probably as a result
of gene duplication and fusion.
For isozyme I, the C-
terminal domain possesses the catalytic activity, and the Nterminal domain is believed to relate to the regulatory
function(8).
On the other hand, in the case of muscle
hexokinase, isozyme II, both the C-terminal and N-terminal
domains exhibit comparable catalytic activity(9,10), i.e..
there are two active sites per monomer.
The crystal structure of yeast hexokinase a and b has
been available for a number of years(11-13).
Mammalian
glucokinase and its glucose binding site has been
modelled(14).
The C-terminal domain of human brain
hexokinase and its ATP binding site has also been modelled
Based on the similarities among the ATP-binding domains of
actin, heat shock protein and glycerol kinase(15).
The ATP
binding site model predicted several amino acid residues(
Thr680, Asp532 and Arg539) interacting with the ATP and sitedirected mutagenesis was performed to confirm the
model(15,16).
In the present study, we examined some other
68
residues that are thought to be located in the putative ATP
binding site.
Experimental Prodedures
Materials—Affi-Gel Blue Gel and Bio-gel hydroxyapatite
were obtained from Bio-Rad Laboratories.
The Transformer™
Site-Directed Mutagenesis Kit(2nd version) was a product of
Clontech.
The Magic Minipreps DNA purification system was a
product of Promega.
Oligonucleotide synthesis and nucleotide
sequence analysis were done by the Iowa State University
Nucleic Acid Facility.
Xho I was from Promega.
from Novagen.
Nru I was from New England Biolabs.
The pET-lla plasmid was obtained
Escherichia coli strains BL21(DE3) and TG-1
cells were obtained from Amersham Corp. E. coli strain ZSC13
is from the Genetic Stock Center, Yale University.
Ampicillin, phenylmethylsulfonyl fluoride(PMSF)*, ATP, NADP,
NADPH, glucose-l,6bisphosphate(Glu-l,6-BP), deoxyribonuclease
I(DNase I), Isopropyl-l-thio-B-D-galactopyranoside(IPTG), and
1,5-anhydro-D-sorbitol were products of Sigma.
Expression of Human Brain Hexokinase—The cDNA for human
brain hexokinase I has been cloned into the expression vector
pET-lla to generate pET-lla-HKI in our lab(17).
pET-lla-HKI
was transformed into E.coli strain ZSC13, which does not
*The
abbreviations
used
are;
PMSF,
phenylmethyl
sulfonylfluoride; IPTG, Isopropyl-B-D-thiogalactopyranoside; Glu
1,6-BP, Glu-l,6-bisphosphate; Glucose-6-P,glucose-6-phosphate; 1,5
an-G6P, 1,5-anhydroglucitol-6-phosphate.
69
contain endogenous hexokinase.
lOO-itiL of preculture of the
transformed E. coli strain in LB medium plus 40mg/L
ampicillin was grown overnight and then added to 10 liter of
the same medium.
The culture was grown in a fermentor at
37°C to early log phase(A6oo=0.4), and IPTG was then added to
a final concentration of 0.4 mM to induce the T7 RNA
polymerase gene.
The culture was grown at a stirring speed
of 200 epm and 5 psi of filtered air flow for 24 hours at
22°C.
Purification of Wild-tpye and Mutant Brain Hexokinase—The
wild-type and mutant forms of hexokinase were purified as
described elsewhere(16).
Site-directed Mutagenesis—Mutagenesis was performed by
following the instructions provided with the Tansformer™
Site-Directed Mutagenesis Kit(2nd version) from Clontech.
The oligonucleotides used for mutagenesis are as follows:
5'-CTGGATCTTGCAGGAACC-3' for Gly534Ala, 5'CTCATTGTTGCGACCGGC-3' for Gly679Ala, 5'-ACCAATTTCMAGTGCTG-3'
for Arg539Lys, and 5^-CAATGCCTGCCACATGGAGG-3 ^ for Tyr686His,
where the underlines represent the mutated bases.
The
selective oligonucleotide sequence was chosen at the Nrul
restriction site in the vector pET-lla and designed to change
the Nrul site to a Xhol site.
The sequence of the selective
oligonucleotide is 5'-CAGCCTCGCCTCGAGAACGCCAG-3'. where the
underline represents the Xhol site.
The DNA sequence of the
mutated gene was verified by DNA sequencing using the method
70
of fluorescent dideoxychain termination sequencing.
Hexokinase Assay—Hexokinase activity was determined
spectrophotometrically as described previously(17).
Preparation of 1.5-an-G6P—1.5-an-G6P was prepared as
described previously(18).
Determination of Protein Concentration—Protein
concentration was determined by the method of Bradford(19),
using bovine serum albumin as a standard.
Circular Dichroism Measurement—Circular Dichroism (CD)
spectra were recorded using a Jasco J710 CD spectrometer as
descrilied elsewhere(16).
Thermostability Experiment—The wild-type and mutant
enzymes were dialyzed against 20 mM of Hepes, containing 1 mM
mercaptoetnanol.
0.25 mg/mL.
The enzyme concentration was adjusted to
The enzyme sample was incubated for 5 minutes at
different temperatures and the residual activity measured.
Fluorescence Emission Spectra—The wild-type and mutant
enzymes were dialyzed against 20 mM of Hepes buffer(pH 7.0),
containing 1 mM of mercaptoethanol.
was adjusted to 50 nq/mL.
The enzyme concentration
Excitation wavelength is 290 nm.
The fluorescent intensity is recorded in the wavelength range
from 300 nm to 350 nm.
71
Results
Purification of the Wild-type and Mutant Human Brain
Hexokinase—The purification procedure for brain hexokinase is
described in the literature(17).
Two methods were modified.
1) E. coli strain ZSC13 was used as the overexpression cell.
E. coli strain ZSC13 is deficient in endogenous hexokinase.
Therefore, contamination by endogenous enzyme activity can be
avoided.
The contamination by endogenous enzyme activity
becomes a serious problem when the mutants exhibit low
hexokinase activity.
2) The fermentor was used to grow cells
and the purification was scaled up.
70 mg of pure enzyme can
be obtained from 20 liter of E. coli cells.
Fig. 4.1. shows
the SDS PAGE obtained with the wild-type and mutant
hexokinases.
Kinetic Studies of the Wild-type and Mutant
Hexokinases-The wild-type and mutant enzymes were studied by
initial rate kinetics and the kinetic parameters are shown in
the Table 4.1.
Gly534 is a conserved residue in actin, heat
shock proteins and sugar kinases, all of which have a common
ATP binding domain(20-23).
Gly534 is in proximity to ATP in
our ATP binding site model.
Gly534 is mutated to Ala.
To test the role of Gly534,
Compared to the wild-type enzyme,
the Gly534Ala mutant showed a 4000-fold decreases in its k^at
value, a 5-fold increase in its K, value for glucose, a 2fold increase in its K, value for ATP, and a 3-fold increase
in its Ki value for glucose-6-P.
These kinetic data suggest
72
1 2 3 4 5 6 7
af
1
1
1
bcd-
—
to
^mm
e-
Fig. 4.1. SDS PAGE patterns for the wild-type and
mutant human brain hexokinase. Lane 2, wild-type, lane 3,
Gly534Ala, lane 4, Gly679Ala, lane 5, Arg539Lys, lane 6,
Tyr686His, Lanes 1 and 7, protein standards; the molecular
masses for the bands a, b, c, d, and e are 200, 116, 97.4,
66.2, and 45 kDa, respectively.
Table 4.1.
A Comparison of the Kinetic data of the wild-type
and the mutant forms of hexokinase
K.(Glu)
K„(ATP)
(MM)
(mM)
K,(l,5-An-G6P)
Hexokinase
± 5.6
(MM)
(s-1)
26 ± 4.8
26.9
wild-type
56
Gly534Ala
287 ± 22.2
2.2 ± 0.13
82 ± 16.5
Gly679Ala
42
± 6.4
0.23 ± 0.39
27 ± 6.0
19.3
Arg539Lys
52 ± 0.39
0.43 ± 0.11
19 ± 4.4
2.2
Tyr686His
43
0.35 ± 0.06
9
±4.1
1.0
± 0.2
± 0.8
0.0063
35.5
The values shown are the mean ± SEM of at least two separate determinations.
kinetic analysis was done with the computer program "Kinetics".
The
74
that the small side chain of Gly534 is essential for
catalysis but it has no significant effect on substrate and
inhibitor binding.
Another conserved residue, Gly679 is also
close to ATP in our ATP binding site model(15).
When Gly679
is mutated to Ala, all the kinetic parameters remained about
the same as that of the wild-type enzyme.
That means that
Gly679 is not essential either for catalysis or binding of
substrates and the inhibitor, glucose-6-P.
Arg539, another
ATP-binding site residue, was mutated to Lys.
The Arg539Lys
mutant exhibited a 12-fold decrease in its
value;
however, little change with xespect to the other kiTietic
parameters was observed.
When Tyr686 was mutated to His, the
koat value and K, values for both glucose and ATP were the
same as those of the wild-type enzyme.
These kinetic data
suggest that Tyr686 is not essential for catalysis ot
substrate binding.
All the wild-type and mutant forms of
hexokinase are inhibited by glucose-6-P and the inhibition
can be relieved by inorganic phosphate(Pi).
These results
suggest that the mutated residues are not important for Pi
regulation.
Effect of Mutations on the CD Spectra—To determine if the
mutations affected the secondary structure of human brain
hexokinase, CD spectra was recorded for the wild-type and the
four mutant enzymes, Gly534Ala, Gly679Ala, Arg539Lys, and
Tyr686His.
The CD spectra were essentially superimposable
for the mutant and wild-type enzymes(data not presented).
75
5
'//y
4
3
i/f I
2
1
Gly534Aia
Gly679Ala
Arg539Lys
0
300
310
320
330
340
wavelength(nm)
Fig. 4.2. Fluorescent emission spectra for wild-type
and mutant forms of human brain hexokinase.
76
suggesting that the mutations did not cause global secondary
structural changes.
Effect of Mutations on the Fluorescent Emission
Spectra—The fluorescent emission spectra were recorded for
the wild-type and all the four hexokinase mutants(Fig. 4.2.).
The Tyr686His mutant spectrum showed some change, suggesting
a structural alteration.
The other mutants, Gly534Ala,
Gly679Ala and Arg539Lys exhibited little change relative to
the spectrum of the wild-type enzyme, suggesting that no
conformation change occurs that might perturb the tryptophan
fluorescence.
Thermostability Experiment—The wild-type and mutant
proteins were incubated at different temperatures for 5
minutes.
The residual activity was determined and the result
are shown in Fig. 4.3.
Only the Tyr686His mutant showed
lower thermostability.
The other three mutants showed the
similar thermostabilities similar to that of the wild-type
enzyme.
Discussion
Fig. 4.4. illustrates the location of the residues that
were mutated in the putative ATP binding site(15).
Gly534 is
only 3.25 A from the oxygen atom of the 6-phosphoryl group of
ATP.
In the Gly534Ala mutant, the hydrogen atom is replaced
with the methyl group of Ala.
This methyl group interacts
sterically with the ATP molecule, causing the large decrease
77
temperature(°C)
Fig. 4.3. Thermostability of the wild-type and mutant
forms of human brain hexokinase. (•)/Wild-type enzyme;
(0),Gly679Ala; (a),Arg539Lys; (•)/ Tyr686His.
78
ATP
ATP
G679
0679
Mg
G534
GLU
G5:
GLU
:539
Fig. 4.4. Stereodrawing of the ATP-binding site of the
model of the ternary complex of brain hexokinase.
79
in catalytic activity but minor changes in the K„ values of
glucose and ATP and the
residue Gly679 is 4.25
value for glucose-6-P.
A
Another
from the oxygen atom of a-phosphoryl
group of the ATP molecule.
When the hydrogen side chain of
Gly679 was replaced with the methyl group of Ala,
the
value for ATP decreased 4-fold while the other kinetic
properties remained the same as those of the wild-type
enzyme.
Probably the larger distance between Gly679 and ATP
permits the substitution of hydrogen by the methyl group
without disturbing the correct orientation of ATP.
The
Gly679Tle mutant was made previously in our lab, but the
Gly679Ile mutant was overexpressed in an unsoluble form.
Probaly the small side chain of G679 is responsible for the
correcting folding of the enzyme.
Arg539 interacts with a-
and 6-phosphoryl groups of ATP in the proposed model,
Arg539
was mutated to Lys to determine the role of the positive
charge of the guanidium group.
The Arg539Lys mutant showed a
small decrease(12-fold) in the keat value,
whereas our
previous(16) results showed that the Arg539Ile mutant
exhibited a 114-fold decrease in the k^at value.
These
results suggest that the positive charge of Arg539 is
important, probably by stabilizing the transition state of
the reaction.
It was shown(22) that the oxygen atom of a particular
water molecule, Wat546 in hsc70, is 3.5A from the y-phosphryl
group of ATP.
Therefore, Wat546 was suggested to attack the
80
y-phosphoryl of ATP in the ATPase reaction of hsc70.
It was
also suggested(ll) that the 6-OH of the glucose analogue No-toluolylglucocamine is in proximity to Wat546.
In hsc70,
Wat546 is polarized by Asp206(Hisl61 in actin), leading to
the suggestion that Asp206 may function as a proton
acceptor.
According to the sequence alignments among actin,
hsc70 and hexokinases from different species, Tyr686 is a
residue corresponding to Wat546.
Tyr686 was mutated to His.
In the present study
The
value of the Tyr686His
mutant and its K. values mutant for both glucose and ATP
were about the same as those of the wild-type enzyme
suggesting that Tyr686 is not essential for catalysis.
This
observation is not surprising because in our model of the
ATP binding site of human brain hexokinase, Tyr686 is far
from the active site.
However, the Tyr686His mutant is less
thermostable. The melting temperature is about 10°C lower
than that of the wild-type enzyme.
It is possible that the
hydrophobic aromatic group of Tyr is important for
stabilizing the correct conformation of human brain
hexokinase.
References
1.
Katzen, H. M. and Schimke, R. T. (1965) Proc. Natl.
Acad. Sci. USA 54, 1218.
2.
Lowry, O.H., and Passonneau, J.V. (1964) J. Biol. Chem.
239, 31.
3.
Rapoport, S. (1968) Essays Biocheiti. 4, 69.
4.
Rudolph, F. B., and Fromm, H. J. (1971) J. Biol. Chem.
246, 6611.
5.
Fromm, H.J. (1981) in The Regulation of Carbohydrate
Formation and Utilization in Mammals (C. M. Veneziale
Ed.), University Park Press, Baltimore, pp45-68.
6.
Ureta, T. (1975) in Isozymes IIIfMarkert. C. L.k ed) pp.
575-601, Academic Press, Inc., New York
7.
Wilson, J. E. (1995) Rev. Physiol. Biochem. Pharmacol.
126, 65.
8.
White, T. K. and Wilson, J. E. (1989) Arch. Biochem.
Biophys. 274, 375.
10. Tsai, H. J. and Wilson, J. E. (1996) Arch. Biochem.
Biophys. 329, 17.
11. Anderson, C. M., Stenkamp, R. E., McDonald, R. C., and
Steitz, T. A. (1978) J. Mol. Biol. 123, 207.
12. Bennett, W. S., Jr., and Steitz, T. A. (1980) J. Mol.
Biol. 140, 211.
13. Harrison, R. (1985) Crystalloaraphic Refinement of Two
Isozymes of Yeast Hexokinase and Relationship of
Structure to Function, Ph. D. Thesis, Yale University,
New Haven, CT.
14. Charles, R. St., Harrison, R. W., Bell, G. I., Pilkis,
S. J., and Weber, I. T. (1994) Diabetes 43, 784.
15. Zeng, C., Aleshin, A. E., Harrison, R. W., Chen, G., and
Fromm, H. J. personal communication.
82
16. Zeng, C., and Fronrni, H. J. (1995) J. Biol. Chem. 270,
10509.
17. Liu, F., Dong, Q., Myers, A. M. and Fromm, H.J. (1991)
Biochem. Biophys. Res. Cominun. 177, 305.
18. Ferrari, R. A., Mandelstam, P., and Crane, R. K. (1959)
Arch Biochem. Biophvs. 80, 372.
19. Bradford, M. M. (1976) Anal. Biochem. 72, 248.
20. Bork, P., Sander, C. and Valencia, A. (1992) Proc. Natl.
Acad. Sci. USA 89, 7290.
21. Kabsch, W., Mannherz, H. G., Suck, D., Pai, E. F. and
Holmes, K. C. (1990) Nature 347, 37.
22. Flaherty, K. M., McKay, D. B., Kabsch, W. and Holmes, K.
C.fl991^ Proc. Natl. Acad. Sci. USA 88, 5041.
23. Kabsch, W., and Holmes, K. C.(1995) FASEB 9, 167.
83
CHAPTER 5. CRYSTALLIZATION AND PREa^IMINARY X-RAY ANALYSIS OF
HUMAN BRAIN HEXOKINASE
A paper submitted to FEES letters for publication
Alexander E. Aleshin, Chenbo Zeng^, Herbert J. Fromm, and
Richard B. Honzatko
Abstract
Human brain hexokinase type I, expressed in Escherichia
coli, has been crystallized from polyethylene glycol 8000 in
the presence of inorganic phosphate. The crystals are hexag­
onal needles of diameter 0.25 mm, diffracting to a resolu­
tion of 3.5
A
on a rotating-anode/area-detector system. The
crystals belong to the space group P3i21/P3221 with cell
dimensions a = b = 171.5
A
and c = 99.4
volume of the crystal is 4.2
A^/Da,
A.
The specific
suggesting an asymmetric
unit with a single 100 kDa molecule and a solvent content of
71% by volume or two molecules of hexokinase with a solvent
content of 41%.
The complex of hexokinase with glucose
crystallizes under similar conditions, giving crystals of
the same morphology.
^Purified the enzyme for crystallization
84
Contents
Hexokinase I (ATP: D-hexose 6-phosphotransferase, EC
2.7.1.1) is one of four mammalian hexokinase isozymes. It is
the pacemaker of glycolysis in brain and the red blood cell
[1,2].
In fact, the human brain uses approximately 50% of
the total chemical energy expended by the human body, a
demand met almost entirely by the activity of hexokinase
type I in the brain [2]. Type I and II isozymes have a
number of features which distinguish them from other
mammalian hexokinases and from
eucaryotic glucokinases,
such as yeast hexokinase. Hexokinases type I and type II
bind to the outer membrane of mitochondria [3] and are
inhibited strongly by their reaction product glucose-6phosphate [1, 4-6].
Hexokinase type I differs from
hexokinase type II in that the product inhibition can be
reversed by physiological levels of inorganic phosphate
[1,4,5].
These properties of hexokinase type I may make it
a critical regulator of glycolysis in brain tissue and
erythrocytes [1,2,4,7].
Human brain hexokinase is a 100-kDa enzyme, consisting
of two highly homologous (50% identity) sequence domains.
Each sequence domain in turn is homologous (30% identity) to
yeast hexokinase and human glucokinase (hexokinase type IV).
Presumably, human brain hexokinase evolved from a 50-kDa
yeast-like enzyme by gene duplication and fusion [8,9].
Hexokinase type I, lacking the C-terminal sequence
85
domain(mini-hexokinase), has the same kinetic properties
toward glucose, ATP, and glucose-6-phosphate as intact brain
hexokinase type I [10].
Mini-hexokinase, however, exhibits
inhibition by glucose-6-phosphate, that cannot be reversed
by phosphate [11,12]. The N-terminal sequence domain is
inactive, generally being regarded as a modulator of
activity [10].
Glucose-6-phosphate binds near or at the ATP
site on the C-terminal sequence domain [1,12], rather than,
as originally suggested [2,13], to an allosteric site on the
N-terminal sequence domain.
On the basis of kinetics [l],
the binding site for glucose-6-phosphate is different from
that of glucose.
The structure of yeast hexokinase is known [14-16], but
the Protein Data Bank [17] has only two relevant entries
with sequences deduced from 2.1
maps.
A
and 3.5
A
electron density
The binding of glucose to the active site causes a
12° rotation of one of the structural domains of the yeast
enzyme, transforming it into an active form [18].
Though
the structure of yeast hexokinase has been used for modeling
the glucose and ATP binding sites of brain hexokinase
[19,20], the yeast enzyme is of little help in understanding
the regulatory properties of hexokinase type I.
Furthermore, yeast hexokinase and hexokinase type I exhibit
significant differences in their interaction
with glucose-6-phosphate [1].
Reported here are the crystallization conditions and
86
preliminary X-ray diffraction investigation of human brain
hexokinase type I.
The cloned enzyme was produced in
Escherichia coli strain BL21(DE3) and purified as described
earlier [12, 21]. The initial conditions for crystallization
were discovered by the method of hanging drops, using a
sparse screening matrix [22].
Crystals grew at room
temperature under the following conditions: 5 nl of the
protein solution of concentration 20 mg/ml in 20-40 mM
sodium phosphate buffer (pH 6.5) was mixed with the same
amount of a precipitant solution containing 3-5%
(w/v)polyethylene glycol 8000 (Sigma) and 0.1 M sodium
cacodylate buffer at pH 6.5. The drops were equilibrated
against 0.7 ml of the precipitant solution.
Long hexagonal
needles of diameter 0.1-0.3 mm appeared within one week.
Crystals of similar morphology also appeared under the same
conditions, but with 10 mM glucose present in the protein
solution. Taking into account that the constants of
dissociation for phosphate and glucose are 22 juM [23] and 60
|iM [12], respectively, we suggest that the first crystal
form contains bound phosphate and the second one both
ligands. The addition of glucose did not change the
morphology of crystals, suggesting that hexokinase type I in
both complexes may have similar conformations.
We collected data from crystals of brain hexokinase type
I, complexed with phosphate, on a Siemens area detector
mounted on an X-ray generator with a Cu rotating anode. We
87
observed diffraction to a resolution of 3.5
A.
Due to
problems of decay, however, we limited data collection to
3.8
A
resolution.
The data set contains 33,375
observations, 15,449 of which are unique reflections (91% of
data expected to 3.8
A).
Data reduction with XENGEN [24]
gave R„g equal to 0.07.
The crystals of hexokinase type I belong to space group
P
3221/P3i21. The unit cell dimensions are
and
c = 99.4
A.
a = b = 171.5
A
If one protein molecule lies in the
asymmetric unit, the V„ is 4.2 AVDa and the fractional
volume occupied by the solvent is 71%. These values are
almost two-fold greater than the average value observed in
other protein crystals [25], which raises the possibility of
two molecules of hexokinase type I in the asymmetric unit
with a solvent content of 41%.
Protein crystals with
trigonal cells, however, generally have a high content of
solvent [25].
The susceptibility of the crystal to damage
by X-rays and its modest diffraction limit, are properties
consistent with a loosely packed asymmetric unit, containing
a single enzyme molecule and a high solvent content.
Acknowledgements:
This work was supported by Grant 95-37500-1926 from the
United States Department of Agriculture and Grant NS 10546
from the National Institutes of Health.
88
References
1. Purich, D. L., Fromiti, H. J. & Rudolph, F. B. (1973)
Adv
Enzymol. Relat. Areas Mol. Biol. 35, 249-326, Wiley
2. Wilson, J. E. (1985) in Reaulation of Carbohidrate
Metabolism (Beitner, R., ed.)» 1/ 45-85, CRC Press, Boca
Raton, FL
3. Gots, R. E., Gorin, F. A., & Bessman, S. P. (1972)
Biochem. Biophvs. Res. Commun. 49, 1249-1255
4. Lowry, O. H. & Passonneau, J.V. (1964) J. Biol. Chem.
239, 31-42
5. Fromm, H. J. (1981) In The Reaulation of Carbohydrate
Formation and Utilization in Mammals (C. M. Veneziale
Ed.), 45-68, University Park Press, Baltimore
6. Ureta, T. (1975) in Isozymes III (C.L. Markert Ed.), 575
601, Academic Press, New York
7. Rapoport, S. (1968) Essavs Biochem. 4, 69-103
8. Easterrby, J. S., & O'Brien, M. J. (1973) Eur. J.
Biochem. 38, 201-211
9. Colowick, S. P. (1973) in The Enzymes (Boyer, P. D.) 3rd
Ed. 9, 1-48, Academic Press, New York
10. White, T. K. & Wilson, J. E. (1989)
Arch. Biochem.
Riophvs. 274, 375-393
11. Magnani, M., Bianchi, M., Casabianca, A., Stocchi, V.,
Daniele, A., Altruda, F., Ferrone, M., & Silengo, L.
(1992)
Biochem. J. 285, 193-199
89
12. Zeng, Ch., & Fromm H. J. (1995) J. Biol. Cheiti. 270,
10509-10513
13. Crane, R. K. & Sols, A. (1954)
J. Biol. Chem. 268, 597-
606
14. Anderson, C. M., Stenkamp, R. E., McDonald, R. C. &
Steitz, T. A.(1978) J. Mol. Biol. 123, 207-219
15. Bennett, W. S., Junior, T. A., & Steitz, T. A. (1980) J.
Mol. Biol. 140, 183-210
16. Harrison, R. (1985) Crystalloaraphic Refinement of Two
Isozymes of veast Hexokinase and Relationship of
Structure to Function. Ph.D. Thesis, Yale University,
New Haven, CT
17. Bernstein, F. C., Koetzle, T. F., Williams, G. J. B.,
Meyer, E. F., Brice, M. D., Rodgers, J. R., Kennard, 0.,
Shimanouchi, T., & Tasumi, T. (1977) J. Mol. Biol. 112,
535-542
18. Bennett, W. S., Junior, T. A., & Steitz, T. A. (1980)
J. Mol. Biol.. 140, 211-230
19. Andreone, T. L., Printz, R. L., Pilkis, S.J., Magnuson,
M. A., & Granner, D. K (1989) J. Biol. Chem. 264, 363369
20. Zeng, C., Aleshin, A. E., Harrison, R. W., Chen, G., &
Fromm, H. J. personal communication.
21. Liu, F., Dong, Q., Myers, A. M., & Fromm, H. J. (1991)
Biochem. Biophys. Res. Commun.. 177, 305-311
90
22. Jancarik, J. & Kim, S. H. (1991) J. Appl. Crvst.. 24,
409-411
23. Ellison, W. R., Lueck, J. D., & Fromm, H. J. (1975) J.
Biol. Chem.. 250, 1864-1871
24. Howard, A., Gililand, G., Finzel, B., Poulos, T.,
Ohlendorf, D., & Salemme, F. (1987) J. Appl.
Crystalloar. 20, 383-387
25. Matthews, B. W. (1986) J. Mol. Biol. 33, 491-493
91
Fig. 5.1. Photograph of human brain hexokinase crystal.
The diameter of the crystal is 0.25 mm.
92
CHAPTER 6. GENERAL CONCLUSIONS
The experiments and results of this study serve to
establish an ATP binding site model for human brain
hexokinase.
This is believed to be the first such model for
mammalian hexokinase.
Based on the similarity of the ATP-
binding domains among actin, heat shock protein, glycerol
kinase and brain hexokinase,
the initial positions and
conformations of MgATP^" complexes in the active site of
brain hexokinase were obtained by superimposing the conserved
residues of the ATP binding site of the enzyme on the
corresponding residues of either actin, heat shock protein or
glycerol kinase, complexed with ligands.
energy minimized.
binding site.
The model was then
Our model predicts the residues in the ATP
Site directed mutagenesis was performed on
those residues to test our model.
Asp532, proposed to bind
to Mg^"^ which forms bonds with 6- and Y~Pl^osphoryl group of
ATP, was mutated to Lys and Glu.
Arg539, which interacts
with the a- and 6- phosphoryl groups of ATP, was mutated to
lie and Lys.
Thr680 which interacts with the y-phosphoryl
group and the oxygen linking the a- and 6-phosphoryl groups
of ATP,
was mutated to Val and Ser.
were determined for the mutants.
The kinetic parameters
The k^,,. values decreased
1000- and 200-fold for the Asp532Lys and Asp532Glu
mutants,respectively; 114- and 12-fold for the Arg539Ile and
Arg539Lys mutants, respectively; and 2000- and 2-fold for the
93
Thr680Val and Thr680Ser mutants, respectively.
The data show
that these residues are important for catalysis.
We suggest
that Asp532 keeps the yphosphoryl group in the correct
position for the attack of 6-hydroxyl group of glucose by
interacting with Mg^^.
We propose that Arg539 stabilizes the
transition state through the interactions of its guanidium
group and the a- and B-phosphoryl groups. In addition, we
suggest that Thr680 keeps the y-phosphoryl group in the
correct orientation for catalysis and also stabilizes the
transition state of the reaction by forming hydrogen bonds
with the Y~phosphoryi group and oxygen which links the a- and
6-phosphoryl groups.
Our model also illustrates several conserved glycine
residues that are located in the ATP binding site: Gly534,
Gly679, and Gly862.
The roles of these glycine residues were
studied by site-directed mutagenesis.
The mutant Gly534Ala
exhibited a 4000-fold decrease in its k^at value.
We suggest
that the small side chain of Gly534 is important so that
enough space is available for ATP binding in the correct
orientation.
The Gly679Ala mutant did not show significant
changes in kinetic parameters, whereas the Gly679Ile mutant
was overexpressed as an insoluble protein.
We suggest that
Gly 679 is important for the correct folding of the enzyme.
The Gly862Ala mutant showed very low activity compared to the
wild-type enzyme, and it did not bind to an Affi-gel Blue-gel
column, a sign of low binding affinity for ATP.
The site-
94
directed mutagenesis studies provide strong support for our
ATP binding site model.
Currently we are preparing mutant enzymes for the
residues involved in adenine binding based on our model.
We
expect that some of them may be responsible for ATP binding,
while others are probably responsible for ATP specificity.
Crystals of human brain hexokinase complexed with
inorganic phosphate, glucose, or glucose-6-phosphate were
obtained.
The ternary complex of human brain hexokinase-ADP-
glucose was also crystallized, and X-ray diffracted data have
been collected.
The solution of the three dimensional
structure is currently being carried out.
The establishment
of the three dimensional structure of human brain hexokinase
will provide important information on how the substrates, ATP
and glucose, bind to the enzyme, and how the inhibitor,
glucose-6-phosphate, and the regulator, inorganic phosphate,
function.
Minihexokinase was overexpressed in the E. coli strain
BL21(DE3) and purified to homogeneity.
The kinetic
parameters were determined for the full-length and mini­
hexokinase.
The K, values for ATP and glucose and the k^^t
values are about the same for both enzymes.
Glucose-6-
phosphate is a competitive inhibitor for both enzymes;
however, the inhibition can be relieved by physiological
concentration of inorganic phosphate only for full-length
hexokinase.
We suggest that the Pi site either resides in
95
the N-terminal half of hexokinase or requires the N-terminal
portion of the enzyme.
To locate the Pi site, Lys290 in the
P-loop sequence was mutated to Ala in full-length hexokinase.
The Lys290Ala mutant did not abolish the enzyme's Pi
deinhibition ability.
for Pi binding.
Therefore Lys290 is not responsible
We plan to do sequential deletion in the N-
terminal portion of the enzyme in order to narrow down the
the location of the Pi binding site.
The purification procedure for human brain hexokinase
was improved, producing lO-fold more protein in half the
time.
The availability of large amounts of the enzyme
improves the possibility that the enzyme can be crystallized
and that it can be studied by NMR spectroscopy.
96
ACKNOWLEDGEMENTS
I would like to use this chance to express my sincere
appreciation to the people who made it possible for me to
accomplish my program of study for my Ph. D. degree, a
milestone of my life.
I would like to thank Dr. Herbert J. Fromm, my major
professor, for his accepting me as a graduate student and for
his wisdom and patience over the years.
I shall always be
grateful to him for his creating an enthusiastic research
atmosphere for everybody in this laboratory.
A very special thanks goes to Alexander E. Aleshin and
Richard B. Honzatko for their excellent cooperation in
molecular modelling and crystallization of human brain
hexokinase.
In addition, thanks go to the visiting
scientists Lirong Chen and Guanjun Chen, and the
undergraduate student Ju-Chi Cheng for their excellent
technical assistance and friendship.
Thanks are also
extended to my committee members for their time and
thoughtful consideration of my research project.
I thank my
fellow graduate students for their encouragement and
friendship, as well as for help with the many laboratory
chores that are part of research.
Finally, I
would like to thank my mother and father,
who raised and nurtured me, always encouraged me to get to
the highest level I can reach, and stood by me faithfully all
97
the way.
I would like to acknowledge my husband and my
daughter for their help, patience and understanding
throughout my graduate education.