.--r

"W

_m<

 

rd‘h Slip

.-‘.- , .31"!
. "Pf-y
w. m.
. H ~t ,.-‘

. .?:;‘v

x ‘5 ‘ -1’ £4“
um'gi'r. ‘r .
‘3. -._€§; ‘2

I

. . .
:35
N J:

-

f. ‘1‘ ‘ 3"" t .
#{éifiUL $5.167 _
l . b 15

.. 1
-., <w

.u

A .1. ;-..n., '

' i '51

I; a?” “ ’ '
' unit. A' ‘ . , ; ‘
-,~. .eszw... --' W” I:
’ ' In x} ’ ' . 43-55 ~ 3:
.‘ ‘ v_. : g .
' '1 '2’ 4‘35, ‘

'h
i

J:

uni
. .-

.4

_.
“1c
"vi
.

$3 “$419.

‘ 've'r ..H- ma.--

 

35. :1.

mfii..:.;.;..;._.'...

 

THESIS

STATE UNIVERSIT

' Milli” llll II III lll’illill

3 1293 01564 9852

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

LIBRARY
Michigan State
University

 

 

 

This is to certify that the

dissertation entitled

Functional Organization and
Evolution of Mammalian Hexokinases

presented by

Henry Jan-Jen Tsai

has been accepted towards fulfillment
of the requirements for

Ph . D . degree in Biochemistry

Major professor

Qazaéai

Date December 3, 1996

 

MS U is an Affirmative Action/Equal Opportunity Institution 0- 12771

PLACE II RETURN BOX to remove thie checkout from your record.
TO AVOID FINES return on or before due due.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU Ie An Afflnnetive Action/Equel Opportunity Inetltuion
Wanna-9.1

 

FUNCTIONAL ORGANIZATION AND EVOLUTION OF MAMMALIAN

HEXOKINASES

BY

Henry Jan-Jen Tsai

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Biochemistry

1996

ABSTRACT
FUNCTIONAL ORGANIZATION AND EVOLUTION OF MAMMALIAN
HEXOKINASES
BY

Henry Jan-Jen Tsai

It has been previously demonstrated that the catalytic
function of mammalian hexokinase I is localized in the C-
terminal half, while the N—terminal half is devoid of any
catalytic activity; It‘was speculated that the active site in
the N-terminal half of hexokinase I has evolved into a
regulatory site. Discovery of a second Glc-6-P binding site
in the C-terminal half of hexokinase I raised the debate
whether the functional regulatory Glc-6-P site is located in
the N- or C- terminal half. By using site-directed
mutagenesis and constructing chimeric hexokinases, the
function of each half of mammalian hexokinases can be examined
and location of the functional Glc-6-P site may be elucidated.

In certain cases, several regulatory functions of
mammalian hexokinases are associated with the N-terminal
halves, e.g. phosphate antagonism of inhibition by AnG-é-P
(1,5-anhydroglucitol 6-phosphate), a glucose 6-phosphate
analogue, correlates with the N-terminal half of hexokinase I
when the C-terminal half is substituted with that of
hexokinase II, and higher sensitivity to phosphate inhibition
correlates with the N-terminal half of hexokinase II when the

C-terminal half is substituted with that of hexokinase I.

Sensitivity to AnG-6-P, however, does not correlate with
either half. In fact, diverse sensitivities are observed in
chimeric hexokinases NICII, NIIICII, NIICI, and NIIICI, which
is perhaps due to the interaction between the two halves.

The unique glucose inhibition associated with hexokinase
III correlates with the C-terminal half. Glucose inhibition
also leads to a lag before attainment of the steady state.

Organization of catalytic function in hexokinase I and
III are similar; the functional active sites are located in
their C-terminal halves. However, hexokinase II has two
functional active sites, one site in each half with similar
specific activities. Because of the dual active sites,
hexokinase II functionally most resembles and is probably the
closest descendant of the ancient product which has arisen
from gene duplication and fusion of a 50 kDa glucose 6-
phosphate sensitive hexokinase.

A new model is proposed for catalysis by the mammalian

type II hexokinase.

Chapter II and III have been reprinted with permission
from .Archives of Biochemistry and Biophysics, copyright
Academic Press, Inc.

Tsai, H.J. and. Wilson, J.E., Functional, Organization of
Mammalian Hexokinases: Characterization of Chimeric
Hexokinases Constructed from the N- and C- Terminal Domains of
the Rat Type I and Type II Isozymes. Arch. Biochem. Biophys.
316, 206-214, 1995.

Tsai, H.J. and. Wilson, J.E., Functional Organization of
Mammalian Hexokinases: Beth N- and C- Terminal Halves of the
Rat Type II Isozyme Possess Catalytic Sites. Arch. Biochem.
Biophys. 329, 17-23, 1996.

Chapter IV has been accepted for publication in.ArChives
of Biochemistry and Biophysics, Academic Press, Inc.

Tsai, H.J. and Wilson, J.E., Functional Organization of‘
Mammalian Hexokinases: Characterization of the Rat Type III
Isozyme and Its Chimeric Forms, Constructed with the N- and
C- Terminal Halves of the Type I and Type II Isozymes. (1996).

TO MY PARENTS

ACKNOWLEDGMENTS

To my parents, is my sincerest appreciation. Earning this
doctoral degree has been the task of my life, and the honor of
receiving this degree should be shared with my parents and
family.

I would also like to extend my appreciation to my Uncle,
Dr. Alan C. Tsai, who always has profound influence in my
life.

My special thank is to Dr. John E. Wilson, who is not
only my dissertation advisor but in many ways has been my role
model, especially as a scientist and scholar.

I would also like to thank my committee, Drs. S.
Ferguson-Miller, L. Kroos, W. Smith, and D. Romsos, who was
the thesis advisor of my Master’ 5 degree at Dept. of Food

Science and Human Nutrition.

vi

Table of Content

List of Tables. ........................................ ...ix
List of Figures.. .................................... . ..... x
Chapter I
Literature Review.. ................ . ..... .............1
Hypothesis and Objective .......................... ...22
References .......................................... .24
Chapter II

Functional Organization of Mammalian Hexokinases:
Characterization of Chimeric Hexokinases Constructed
from the N- and C- Terminal Domains of the Rat Type I
and Type II Isozymes ................................28

Material and Methods..... ............. ...............34
Results ............................................ ..41
Discussion .................................. - ....... ..60
References ........................................... 64

Chapter III

Functional Organization of Mammalian Hexokinases:
Both N- and C- Terminal Halves of the Rat Type II
Isozyme Possess Catalytic Sites .....................69

Material and Methods ............................... ..74

Results... ......................................... ..78

Discussion.... ........... ‘ .......................... ..90

References.... ....................................... 98
Chapter IV

Functional Organization of Mammalian Hexokinases:
Characterization of the Rat Type III Isozyme and Its
Chimeric Forms, Constructed with the N- and C- Terminal
Halves of the Type I and Type II Isozymes...........101

Material and Methods ................................ 106

vii

Results....... ..................................... .114
Discussion. ..... . .................................. .130

ReferenceSOOOOOOOOOO0.0.0.000...0.0.0000000000000000140

Chapter V
Conclusion...................... .................. ..144
Future Research ..................................... 154
References ................................ . ..... ....156

viii

LI ST OF TABLES
Chapter II
Table I. Hexokinase activity in extracts of COS-1 cells
transfected with cDNAs for type I or type II hexokinase
or for chimeric hexokinases..... ..... . ........... ....44

Table II. Kinetic properties of type I and type II
hexokinase and the chimeric hexokinases .............. 54

Chapter III
Table I. Hexokinase activity in extracts of COS-1 cells
transfected with cDNAs encoding wild type and mutant
forms of type II hexokinase ......... . ..... ...........80

Table II. Kinetic properties of wild type and mutant
forms of type II hexokinase .............. .. .......... 86

Table III. Hexokinase activity in extracts from COS-1
cells transfected with cDNAs encoding wild type and

chimeric forms of hexokinase ......................... 91
Chapter IV

Table I. Hexokinase activity in extracts of transfected

COS-1 and M+R 42 cells .............................. 115

Table II. Km'Values for glucose and ATP ............. 121

ix

LIST OF FIGURES
Chapter II
Figure 1. Nucleotide and deduced amino acid sequences
surrounding the centrally located Nco I sites in cDNAs

for rat type I and type II hexokinases...............36

Figure 2. Restriction analysis of parental cDNAs and
chimeric constructs .......................... ........42

Figure 3. SDS gel electrophoretic analysis of COS-1
cell extracts....... ...... ...........................46

Figure 4. Western blot analysis of COS-1 cell extracts
... ............ .. ............................. .......48

Figure 5. Analysis of COS-1 cell extracts by
isoelectric focusing..... ............ ......... ....... 50

Figure 6. Analysis of COS-1 cell extracts by
electrophoresis under nondenaturing conditions.. ..... 51

Figure 7. Heat stability of type I, type II, and
chimeric hexokinases ....... . ..... ....................53

Figure 8. Inhibition of type I, type II, and chimeric
hexokinasesby Glc-1,6-bisphosphate ................. .56

Figure 9. Inhibition of type I, type II, and chimeric
hexokinases by Pi .................................. ..58

Figure 10. Effectiveness of Pias an antagonist of
inhibition by AnGlc-6-P. ....... .... ..... ..... ........ 59

Chapter III

Figure 1. SDS—gel electrophoretic analysis of COS-1
cell extracts............... ..... ....................81

Figure 2. Immunoblot analysis of COS-1 cell extracts

......OOOO0.0.0.0..........OOOOOOOOOOOOO00.0.0000000083

Figure 3. Densitometric analysis of an Immunoblot

......OOOOOOOOOOOOOOO......OOOOO0.00.00.00.000000000084

Figure 4. Inhibition of wild type and mutant forms of
type II hexokinase by the Glc-6-P analog, 1,5-
anhydroglucitol-G-P ............ ....... ....... ........87

Figure 5. Inhibition of wild type and mutant forms of
type II hexokinase by inorganic phosphate............88

Figure 6. Evolution of the isozymes of mammalian
hexokinase .......................... . ........ . ....... 95

Chapter IV

Figure 1. Characterization of expressed hexokinase
isozymes and chimeric forms by immunoreactivity.....116

Figure 2. SDS gel electrophoretic analysis of M+R 42
cell extracts ........................... . .......... .118

Figure 3. Electrophoresis of expressed isozymes or
chimeric forms under nondenaturing conditions.......120

Figure 4. Substrate inhibition of the type III isozyme
and the NICIII and NII’CIII chimeras by glucose.....123

Figure 5. Inhibition by 1,5-AnGlc-6-P at 3.3 mM Glc

00000000000000000 0000000000000 00000 ....... 0000000000125

Figure 6. Inhibition of type III hexokinase and the
NICIII and NII'CIII chimera by 1,5-AnGlc-6-P at 0.5
mMG1C0000000000000000 00000 0000000000000000000000000127

Figure 7. Inhibition by Piat 3.3 mM Glc

000 00000 0000000000 00000 0000 0000000000 0 000000 00000000129

Figure 8. Lack of an effect of Glc concentration on the

inhibition of type III hexokinase and the NICIII and

NII'CIII chimeras by Pi ................ .............131
Chapter V

Figure 1. Proposed symmetric model for catalysis by
mammalian type II hexokinase.... ........ ............150

xi

Chapter I

LITERATURE REVIEW

2

The reaction catalyzed by hexokinases

Hexokinase (ATP: D-hexose 6-phosphotransferase (EC
2.7.1.1) is the first enzyme in glycolysis, which catalyzes
the following reaction: .

Glc + ATP.Mg“ ------- > Glc-6-P + 111313.1492+

This reaction is the first reaction after glucose being
transported. into cytoplasmic space, and is, in. essence,
phosphorylation of glucose at the expense of high energy ATP
(AG = -7.3 kcal/mole hydrolyzed at pH 7). The reaction forms
a lower energy phosphoglucose, Glc-6-P (AG = -3.3 kcal/mole
hydrolyzed at pH 7); therefore, the AG for this reaction is

-4.0 kcal/mole at pH 7 (1). This reaction not only activates
glucose for further metabolism but because of the large
negative free energy change, the reaction is thermodynamically
favored to the right and essentially irreversible (2).

The product of this reaction, Glc-6-P, is a common
substrate for phosphohexoisomerase, phosphoglucomutase, and
Glc-6-P dehydrogenase, which introduce Glc-6-P into
glycolysis, glycogen synthesis, and the pentose phosphate
pathway, respectively. Thus, hexokinase plays an important
role in admitting glucose into metabolism.

As the name implies, hexokinases are able to
phosphorylate a variety of hexoses (3), e.g. mannose, 2-
deoxyglucose, glucosamine and, to a lesser extent, fructose
and galactose. .All of these hexoses are phosphorylated at the

6-hydroxyl position, forming hexose 6-phosphate.

3
Evolution of mammalian hexokinases

There are four distinct hexokinase isozymes found in
mammalian tissues (3-5), designated as type I, II, III, and IV
hexokinase. Type I, II, and III isozymes are monomeric with
molecular weight of approximately 100 kDa, whereas type IV
isozyme, better known as glucokinase, is only 50 kDa.

Because the molecular weight of mammalian hexokinase I,
II, and III is twice that of yeast hexokinase (50 kDa), it has
been suggested first by Easterby (6) and Colowick (7) that
mammalian loo-kDa hexokinases arose by duplication and fusion
of a gene encoding an ancestral 50 kDa hexokinase similar to
present-day yeast hexokinase. If this hypothesis is true, the
original ancestral 100 kDa hexokinase should have two
identical halves with one active site in each half (6,7).
Easterby and Colowick further postulated that this original
100 kDa hexokinase in turn gave rise to modern hexokinase I,
II, and III.

By comparing amino acid sequences deduced from cDNAs of
various sources, hexokinase I, II, and III are indeed very
similar (5,8-10). In addition, they exhibit internal 50 kDa
repetitions which are similar to the sequence encoding.50 kDa
yeast hexokinase (5,8-10) . Furthermore, several residues
involved in glucose binding or catalysis are well conserved in
each repetition as well as in yeast hexokinase (5,8-10),
supporting the theory that modern mammalian 100 kDa

hexokinases are derived from a common ancestral hexokinase

4
which itself was a product of gene duplication and fusion
(5,11,12).

One important piece of evidence supporting this gene
duplication and fusion theory is by Printz et al. (13) and
Kogure et a1. (14) in which they found that the splicing sites
and the exon sizes of hexokinase II gene showed a direct
repetition between the N- and C- terminal halves.
Interestingly, this intron/exon pattern is also observed in
glucokinase:gene (13,14), and therefore they suggested that an
ancestral gene encoding a 50 kDa hexokinase similar to
glucokinase underwent gene duplication and fusion to form the
hexokinase II gene.

However, instead of glucokinase being the precursor of
hexokinase II, Griffin et a1. (12) suggested an alternative
view'in which.the glucokinase gene arose by resplitting of the
hexokinase II gene and thus it retained the intron/exon
pattern of the hexokinase II gene. In fact, the latter view
is perhaps a better model, because it is better supported by
the evolution scheme. In lampreys (Agnatha; jawless fish),
only one hexokinase isozyme is present, which is about 90 kDa
in size (15), while type I, II, III isozymes but not
glucokinase are present in further evolved Osteichthys (bony
fish). Glucokinase is not present until the emergence of
Amphibia and higher animals (except for Aves; no glucokinase
is found in Aves) (15,16). These findings would suggest that

the gene duplication and fusion event of hexokinase occurred

5
by the time when primitive fish emerged, while glucokinase did
not arise until the emergence of Amphibia. Furthermore,
Griffin et a1. (12) found that glucokinase has a closer
sequence similarity to the mammalian 100 kDa hexokinases than
to yeast hexokinase. Therefore, the glucokinase gene seems
more likely to be a product of the splitting of the hexokinase

II gene, not the precursor.

Properties of mammalian hexokinases

Mammalian hexokinases I, II, and III are sometimes
referred as low'Km hexokinases (4,5), because they have low
Km, i.e. high affinity, for glucose (Km: 10-150 uM).
Mammalian low lg, hexokinases are sensitive to inhibition by
the product, G1c-6-P, with Ki being about 10-100 uM. Type IV
isozyme, on the other hand, has lower affinity for glucose
with K, being about 5-10 mM, and unlike low Km hexokinases, it
is not sensitive to inhibition by physiological concentrations
of Glc-6-P (3,5).

While hexokinase I is the dominant isozyme found in most
tissues, hexokinase II is mostly found in those insulin
sensitive organs and tissues, e.g. muscles, liver, and adipose
tissues, due to, partially at least, the up-regulation of
hexokinase II by insulin (13,17). Because of this tissue
distribution and hormonal regulation along with other kinetic
properties, it has been speculated that the hexokinase I is

responsible for introducing glucose into glycolysis, while

6
hexokinase II is responsible for converting glucose into
glycogen in insulin sensitive organs and tissues (4,18) (also
see Kinetics of hexokinases).

Intracellular localization of hexokinases is not
homogeneous (17), which is believed to bear functional
significance (5). A large portion of hexokinase I and II was
found to co-sediment with mitochondria (17). Later,
hexokinase I and II were shown to be bound to the
mitochondrial membrane (4,5,17) by interaction of the N-
terminal hydrophobic sequences and, presumably, mitochondrial
membrane lipids (19) and by interaction between the negative
charge on hexokinase surface and presumably mitochondrial
membrane phospholipids, bridged by divalent cations (20).

In the case of brain hexokinase I, Glc-6-P is able to
cause dissociation of hexokinase I from mitochondria to
various extent depending on species; the degree of
dissociation ranges from 90% in rats to 20% in human, with
guinea pig and bovine in between (21). Whether hexokinase II
has the same properties is not known. It is noteworthy that
there persists a portion of bound hexokinase resistant to Glc-
6-P solubilization, which can be, however, solubilized by 0.5
M KSCN. Comparison of the Glc-6-P soluble and KSCN-
solubilized.hexokinases showed.no»distinguishable difference,
which led to the postulation by Kabir and Wilson (21) that
there are two subtypes of hexokinase—binding-sites on the

mitochondrial surface. That is, solubilities of the

7
'hexokinase bound on mitochondria are determined by the nature
of hexokinase-binding-sites rather than by different subtypes
of hexokinase or mitochondria.

The functional significance of intracellular localization
of hexokinase I on mitochondria is suggested by BeltrandelRio
and. Wilson (22) as the feedback link between oxidative
phosphorylation and glycolysis. Because of the close
proximity, mitochondria-bound hexokinase I utilizes ATP
primarily generated from mitochondria to phosphorylate glucose
(22). In this scenario, when the energy charge state in the
cell becomes low and mitochondrial oxidative phosphorylation
is stimulated, the increased flux of ATP from mitochondria
would elevate the flux of glucose phosphorylation (by
hexokinase I) which in turn supports the need for carbon
sources in glycolysis and Krebs cycle.

Hexokinase III is found in many tissues but in much lower
amount than other hexokinase isozymes (23). Intracellular
localization of hexokinase III was thought to be in the
cytoplasmic space, but.recently hexokinase III was found.to be
localized at the nuclear periphery. The association of
hexokinase III to the nuclear periphery is apparently rather
weak, and.disrupted.when tissue is homogenized (23). The role
of hexokinase III and its intracellular localization in cell
physiology is not clear.

Glucokinase is mainly found in liver (24, 25) and in

pancreas (25). Because its Km for glucose is in the range of

8

circulating blood glucose levels, fluctuation of blood glucose
can greatly influence the rate of glucose phosphorylation
catalyzed by glucokinase. During a postmeal high blood
glucose state, the high Km for glucose makes glucokinase ideal
to convert excess blood glucose into glycogen in liver or to
serve as a sensor governing the release of insulin in the
pancreatic B-islets. A

Glucokinase was thought to be a cytosolic enzyme;
however, Miwa et a1. (26) have demonstrated that glucokinase
is present in the nucleus as well as in the cytosol. The
physiological significance of this intracellular localization

of glucokinase is not known.

Kinetics of hexokinases

The hexokinase reaction proceeds by a ternary complex
mechanism, i.e. hexokinase-glucose-ATP.Mg2+ must be formed
before the catalysis can occur (4). Whether the substrate
binding follows a sequential order or random order has been
the subject of a long debate (4,5). Fromm and his coworkers
reported that the binding order is a rapid equilibrium random
mechanism (18,27,28); that is, either glucose or ATP.Mg“'can
bind to hexokinase first, followed by the binding of the other
substrate. However, most other investigators reported that
substrate binding follows a sequential order in which glucose
binds first (4,5,7). To prove the latter view and to avoid

the inherent. ambiguity] of 'traditional product inhibition

9

kinetics, Gregoriou et al. (29) demonstrated, with the flux
ratio method, that the substrate binding is a sequential
ordered mechanism with glucose binding to the enzyme first.
Later Ganson and Fromm (30) rebutted with results using the
equilibrium isotope exchange method, and reported that the
reaction operates by a random mechanism but with a preferred
pathway, in which two thirds of the reaction proceeds with
glucose binding first and.ATP.Mg“’later. Since the mid 1980,
the argument about the kinetic mechanism of hexokinases seems
to have subsided, and to this date, there is no clear and
decisive report on whether hexokinases follow a sequential
order or a random order in the substrate binding.

Glc-6-P is a competitive inhibitor vs. ATP.Mg“’(3-5,7),
indicating Glc-6-P and ATP.Mg“ are mutually exclusive when
binding to hexokinase. Whether Glc-6-P binds at a distinct
allosteric site or a site overlapping with the ATP.Mg2+ is
another ongoing debate (see Allosteric inhibition of
hexokinase and Functional organization of hexokinases in this
chapter).

Most investigators ihave reported. that. Glc-6-P is a
noncompetitive inhibitor vs. glucose (3-5,31,32). However,

their double reciprocal (Lineweaver-Burk) plots showed almost

parallel lines (31,32), which. is characteristic of
uncompetitive inhibition. In fact, Fromm and Zewe (33)
reported that the inhibition is uncompetitive. This near

parallel pattern, or near uncompetitive inhibition, is

10
probably due to the synergistic effect of glucose and Glc-6-P
binding (34,35), in' which the converging point of the
Lineweaver-Burk plot becomes distant from the origin (36).
Thus, this unusual "mixed type" inhibition pattern closely
resembles uncompetitive inhibition.

One other.unique characteristic of hexokinase I is that
inorganic phosphate can antagonize Glc-6-P inhibition at low
mM concentrations (11,27,28,37,38). This unique
characteristic is believed to play a significant role under
certain physiological conditions in the brain (11). In brain,
hexokinase I is considerably inhibited by its product, Glc-6-
P, under normal conditions. When an energy demanding
circumstance arises, the elevated phosphate concentration,
resulted from hydrolysis of ATP, serves to relieve hexokinase
I from Glc-6—P inhibition (11) “and thus hexokinase I
introduces more glucose into glycolysis and-the Krebs cycle to
meet the energy demand. At higher concentrations (>5 mM),
however, phosphate starts to exhibit its inhibitory effect on
hexokinase I (4,5,38).

In the case of hexokinase II, antagonism of Glc-6-P
inhibition by inorganic phosphate is not seen and.phosphate is
solely an inhibitor with Ki being about 3 mM (18,38). This
inhibitory effect of inorganic phosphate would suggest that
during muscle exercise, elevated phosphate concentrations,
again resulting from ATP hydrolysis, could inhibit hexokinase

II and at the same time activate glycogen phosphorylase and

11

phosphofructokinase (18). As a result, myoglycogen is
funneled into glycolysis to generate more ATP for muscle
contractions. Such a mechanism indicates that the source of
carbon fuel for muscle contraction is myoglycogen not free
glucose; therefore, hexokinase II is responsible for
converting glucose to myoglycogen rather than introducing
glucose into glycolysis.

Compared to hexokinase I and II, hexokinase III has lower
Kfi for glucose (10-30 uM) and higher KifOr Glc-6-P (approx.
loo‘uM) (3-5), and uniquely, hexokinase III exhibits substrate
inhibition by glucose in the mid to high mM range (3,39). The
physiological significance of these kinetic properties is not

clear.

Allosteric inhibition of hexokinases

By using a series of analogues, Sols and Crane (40,41)
discovered that brain hexokinase I has different specificities
for the hexose moiety of substrates and inhibitors, and
therefore suggested that the inhibitor occupies a distinct
allosteric site' different from the active site where
substrates bind. For example, both mannose and 2-deoxyglucose
are good substrates for hexokinase I, but their phosphorylated
counterparts, mannose-6-P and 2-deoxyglucose-6-P are poor
inhibitors. This hypothesis was further tested by Sols (42)
and Ureta et al. (43) with the reverse reaction catalyzed by

hexokinase I, in which G1c-6-P serves both as a substrate and

12

as an inhibitor. The reverse reaction showed substrate
inhibition when Glc-6-P was used as the substrate, whereas 2-
deoxyglucose-6-P showed no substrate inhibition due to its
poor ability to inhibit hexokinase I. Furthermore, yeast
hexokinase, which is insensitive to Glc-6-P inhibition, showed
no substrate inhibition when Glc-6-P was used as the substrate
in the reverse reaction. These experiments (43) suggested
that G1c-6—P exerts it inhibitory effect on hexokinase I by
binding to a distinct allosteric site rather than the active
site.

The separate allosteric site in hexokinase I is also
evident from the fact that Glc-6-P is not a competitive
inhibitor vs. glucose (3). If Glc-6-P inhibits hexokinase I
by binding to the active site, it should have been a
competitive inhibitor vs. glucose, since both molecules cannot
occupy the same active site at the same time. The fact that
Glc-6-P is not a competitive inhibitor vs. glucose and that
both G1c-6—P and glucose can bind to hexokinase I
simultaneously (34,35) confirms the discrete allosteric and
active sites on hexokinase I. In fact, Glc-6-P and glucose
showed synergism when they bind to hexokinase I (34,35); that
is, binding of either ligand will promote the binding of the
other ligand. Thus, Glc-6-P must bind at a discrete
allosteric site, which does not overlap with the active site

where glucose binds.

13

Structure and ligand binding

Although there is no mammalian 100 kDa hexokinase crystal
structure available, a significantly related hexokinase from
yeast (50 kDa) has been<crystallized.and.structures.determined
at 2.1 A and 3.5 A resolution (44-47). Since mammalian 100
kDa hexokinases are believed to arise from duplication and
fusion of a gene encoding a 50 kDa hexokinase similar to the
yeast hexokinase (5-1’4) , these 50 kDa yeast hexokinase crystal
structures have been used for modeling mammalian 100 kDa
hexokinases (48,49) as well as 50 kDa glucokinase (50-51).

These two yeast hexokinase structures represent two
different states during hexokinase catalysis: one is in the
"open"-conformation (resolution at 2.1 A), which represents
the structure before the glucose-induced cleft closure; the
other is in the "closed" conformation (resolution at 3.5 A)
with a glucose molecule trapped at the active site. The
enzyme consists of two lobes surrounding the active site, a
structural feature also found in many other kinases (45).
When«glucose binds to the active site between the two lobes of
the enzyme, the smaller lobe of the enzyme closes down and
bends 12° relative to the large lobe, forming the "closed"
conformation (45-47). The enclosure of glucose in the active
pocket effectively shields most of the glucose molecule from
the solvent environment, exposing only the 6-hydroxyl group
for accepting the phosphate transfer from ATP.

From these structures, it is observed that Ser 158, Asn

14

210, Asp 211, Gln 269, and Gln 302 are involved in hydrogen
bonding to the 1-, 3-, 4-, and 6- hydroxyl groups on the
"trapped" glucose molecule (46,47). Interestingly, these
hydroxyl groups have been reported by Crane and Sols (40,41)
to be critical for hexose specificity in brain hexokinase I.
Such observation is probably not coincidence, but rather it
reinforces the concept that mammalian hexokinases arose from
an ancestral hexokinase similar to present-day yeast
hexokinase.

Since the active site of hexokinase I is located in the
C-terminal half (11,49,53-56) , one may expect that alterations
in one of these corresponding amino acid residues in the C-
terminal half may affect glucose binding of hexokinase I.
Later mutageneses studies (49,53-55) were consistent in
general with this hypothesis; alterations in one of those
corresponding residues (Asp 657, Gln 708, Gln 742) in the C-
terminal half of hexokinase I hindered the glucose binding
affinity and thus catalytic activity of hexokinase I, while
alterations in those corresponding residues in the N-terminal
half have no such effect.

One rather surprising observation made by Arora et al.
(50) was that mutation of $603A in the C-terminal half, as one
may expect, decreased the catalytic activity of hexokinase I,
but the binding affinity for glucose was surprisingly
increased. Although Harrison (47) suggested that the 3-

hydroxyl group of glucose forms hydrogen bond with the oxygen

15
on the carbonyl group, not the hydroxyl group on the side
chain, of corresponding Serine 158 in yeast hexokinase, how
the substitution of Ser to Ala would actually increase the

glucose binding affinity remains to be examined further.

Functional organization of hexokinases
Easterby (6) and.Colowick (7) first hypothesized.that.the

 

allosteric site of hexokinase I evolved from one of the two
active sites, leaving the other active site intact for
catalysis. This hypothesis was later supported by ligand-
protected limited. proteolysis of rat brain. hexokinase I
(11,56). White and Wilson demonstrated that Glc-6-P was able
to protect the N-terminal half of hexokinase I from trypsin
attack under mild denaturing conditions, while the glucose
analogue, N-acetyl-glucosamine, along with ATP were able to
protect the C-terminal half under the same conditions.
According to the hypothesis, one may expect that N- and C-
terminal halves of hexokinase I isolated under such conditions
will retain their specialized function; that is, the N-
terminal half will show no catalytic activity, but C-terminal
half will retain full catalytic activity with no sensitivity
to Glc-6-P inhibition.

Surprisingly however, the C-terminal half prepared under
such conditions not only possesses full catalytic activity of
hexokinase I, but is also sensitive Glc-6-P inhibition, which

implies that the C-terminal half possesses its own Glc-6-P

16

binding site (11). Upon further examination, the N-terminal
half of hexokinase I‘was found to have its own.glucose binding
site as well as a Glc-6-P binding site. In other words, both
N- and C—terminal halves have a glucose binding site and a
Glc-6-P binding site; therefore, White and Wilson (11)
suggested that the ancestral 50 kDa hexokinase 'must have
acquired an additional allosteric site before the gene
duplication and fusion event occurred. Indeed, such 50 kDa
Glc-6-P sensitive hexokinases have been found in silkworm
(57), in the parasite, Schistosoma mansoni (58), in starfish
(11,59) and other marine species. Hence the allosteric site
in hexokinase I did not evolve from an active site but rather
was shaped by nature before the gene duplication and fusion
event occurred.

Although there are potentially two active sites and two
allosteric sites present in each 100 kDa hexokinase, Chou and
Wilson (60) found that there is only one glucose binding
(active) site. and. one: Glc-6-P binding (allosteric) site
detectable.by Scatchard analysis in each 100 kDa hexokinase I.
Similar observations have also been reported by Ellison et al.
(34). Since Scatchard analysis :measures the functional
binding sites, White and Wilson (11) suggested that the
"active site" in the N-terminal half and the "allosteric site"
in.C-terminal half of hexokinase:I are latent and inaccessible
to substrate or inhibitor, and therefore they are not

detectable by Scatchard analysis.

17

Since the catalytic activity' has been shown to be
associated with the C-terminal half of hexokinase I by several
laboratories (11,49,53-56), it is obvious that the functional
active site of the intact hexokinase I is located in the C-
terminal.

The location of the functional Glc-6-P binding site,
whether it is located in the N- or C-terminal half of
hexokinase I, remains in dispute (11,49,53-56). However,
since Glc-6—P can protect the N-terminal half of hexokinase I
from trypsin digestion (11,56) or sulfhydryl modification (61)
at the uM levels, which are comparable to the Kivalue of Glc-
6-P, White and Wilson (11) suggested that the functional Glc-
6-P binding site is located in the N-terminal half.

Several other investigators (49,62,63) have suggested
that the functional Glc-6-P binding site is located near the
active site in the C-terminal half rather than in the N-
terminal half. .Arora et al. (49) reported that the functional
Glc-6-P site is the one in the C-terminal half because 1)
truncated 50 kDa C-terminal half of hexokinase I is sensitive
to Glc-6-P inhibition, and 2) mutation of residues in the N-»
terminal half corresponding to those responsible for glucose
binding in the C-terminal half of hexokinase I has no effect
on the Ki of G1c-6-P.

Nevertheless, these arguments are not flawless. First,
glucose and Glc-6-P binding sites in both N- and C-terminal

halves are accessible in the isolated 50 kDa truncated forms.

18

In the intact 100 kDa form, however, only one glucose site and
one Glc-6-P site are accessible in hexokinase I (34,60). The
fact that Glc-6-P inhibits the 50 kDa C-terminal half does not
necessarily mean this Glc—6-P site in the C-terminal half is
the regulatory site in the 100 kDa form. If their argument is
correct that Glc-6-P binds, at the C-terminal half of
hexokinase I, then why is the N-terminal half protected by
Glc-6-P from proteolysis? If Glc-6-P can bind at both the N-
and C-terminal halves, then. why' does Scatchard analysis
indicate there is only one Glc-6-P binding site in the entire
hexokinase I?

Second, though Solheim and Fromm (62) and Jarori et al.
(63) have suggested that Glc-6-P binds very close to the
active site (glucose binding site), the fact that Glc-6-P and
glucose can bind to hexokinase I simultaneously (34,35) and
that Glc-6-P is not a competitive inhibitor vs. glucose even
with the isolated 50 kDa C-terminal half (11), suggests that
there is no overlap between the Glc-6-P binding site and the
glucose binding site. Therefore, it is very unlikely that
both.glucose and.Glc-6-P share several residues for binding to
the enzyme. That is, residues responsible for glucose binding
are not necessarily responsible for Glc-6-P binding. Thus,
mutation at residues responsible for glucose binding in the N-
terminal half of hexokinase I may have little, if any, effect
on Glc-6-P binding (49), and mutation of such residues does

not provide information about the location of the functional

19
allosteric site.

Jarori et al.(63) reported that glucose and Glc-6-P are
11 A and 8 A, respectively, away from a Mn“ binding site,
which means that glucose and Glc-6-P may be as Close as 3 A
and as far as 19 A from each other when bound to hexokinase I.
Considering that the diameter of a glucose is no more than 6
A (distance between the oxygen atoms on carbon 2 and 6), and
a typical hydrogen bond is in the range of 2.6 - 3.1 A
(distance between the donor atom and the acceptor atom), the
minimal 3 A distance between glucose and Glc-6-P still
represents a relatively large distance, and such distance
would seemltijreclude the possibility that glucose and.Glc-6-
P share the same residues for binding (see the previous
paragraph). The actual distance between glucose and Glc-6-P
is not known, but likely to be more than 3 A.‘ Based on the
result of Jarori et al.(63), whether Glc-6-P binds close to or
far from glucose is a matter of individual belief and
perception.

Solheim and Fromm (62) determined the Km of Glc-6-P in
the reverse reaction of hexokinase I to be about 25 uM.
Because of this strong affinity, Fromm insisted that this is
the high affinity site for Glc-6-P and it represents the
inhibitory nature of Glc-6-P, i.e. Glc-6-P functions as an
inhibitor by binding to the active site with a high affinity.
However, this speculation conflicts with one important fact.

The fact that Glc-6-P is not a competitive inhibitor vs.

20
glucose and both molecules can bind to hexokinase I
simultaneously precludes the possibility that the inhibitory
Glc-6-P binds at the active site (see Allosteric inhibition of
hexokinases).

Solheim and Fromm (62) then proposed.a model in which the
Glc-6-P binding site is located next to the active site. In
this model, the phosphate moiety'of Glc-6-P occupies the space
of the 7-phosphate of ATP, which explains the competitive
inhibition nature of Glc-6-P vs. ATP. The glucose moiety of
Glc-6-P, however, occupies a distinct site from the glucose
binding site, which explains why glucose and Glc-6-P can bind
to hexokinase simultaneously. They proposed that when Glc-6-P
is formed after the catalysis, the phosphate moiety remains in
place but the glucose moiety dissociates and binds to a new
site. Note that in this model, the site for binding to the
glucose moiety of Glc-6-P is distinct from the active site
(glucose binding site).

However, Fromm’s explanation for how'Glc-6-P binds to its
new site after catalysis is not well supported by structural
observations from Steitz (45). Glucose has a strong affinity
for hexokinase with Km of about 50 uM and the ability to
induce a conformational change in hexokinaseI (35,45-47).
Based on the crystal structure of yeast hexokinase, Steitz
speculated that because of the glucose induced conformation
change, the dissociation of Glc-6-P is perhaps the rate

limiting step in. hexokinase catalysis (45), where cleft

21
opening is the limiting factor to Glc-6-P dissociation. In
such.case, it would.be hard to conceptualize hOW'the phosphate
moiety remained anchored to the enzyme, while the glucose
moiety would dissociate from the enzyme and bind to a

different site.

HYPOTHESIS AND OBJECTIVE

To find out whether the locationlof the regulatory Glc-6-
P site is in the N- or the C-terminal half of hexokinase I,
chimeric hexokinases constructed with the N- and C-terminal
halves from type I, II, and III isozymes were employed. The
hypothesis of this study is that Glc-6-P binds to the N-
terminal half of hexokinase I, and therefore, the regulatory
function should be associated with the N-terminal’ of
hexokinases.

This work was initiated to test such hypothesis.
However, during the course of this study, it became evident
that the functional organization in hexokinase II is quite
different from those in hexokinase I and III (64,65). This
observation has an impact on the current evolution scheme
proposed by White and Wilson (11) and Griffin et al. (12);
subsequently, the evolution scheme of mammalian hexokinases
was revised to accommodate such observation (64).

In addition to constructing chimeric hexokinases from
hexokinase I and II, modification of hexokinase III cDNA was
needed. The cDNA of hexokinase III has been previously
subcloned as two separate pieces with several undesirable
restriction sites in the open reading frame (65); therefore,
assembly and extensive site-directed mutageneses were needed
to bring the hexokinase III cDNA into a form that is
compatible with type I and II cDNAs (65). Completion of this

task would not.only'allow'construction.of chimeric hexokinases

22

23
with type III isozyme, but also provide an opportunity to
examine its kinetic properties. Among those hexokinase III
properties, glucose inhibition is unique and particularly
interesting. Thus, it is of great interest to determine the
half/domain that is responsible for this glucose inhibition as

well as the overall functional organization in hexokinase III.

100

11.

12.

13.

14.

REFERENCES

Conn, E.E., and Stumpt, P.K. (1976) Outlines of
Biochemistry, 4th ed., John Wiley, New York.

Siegel, G.J., Agranoff, B.W., Albers, R.W., and Molinoff,
P.B. (1989) Basic Neurochemistry, 4th ed., Raven Press, New
York.

Grossbard, L., and Schimke, R.T. (1966) J. Biol. Chem. 241,
3546-3560.

Wilson, J.E. (1984) in Regulation of Carbohydrate
Metabolism (Beitner R., ed.) Vol. I, pp45-85, CRC Press,
Boca Raton, FL.

Wilson, J.E. (1995) Revs. Physiol. Biochem. Pharmacol. 126,
65-198.

Easterby, J.S. (1971) FEBS Lett. 18, 23.

Colowick, S.P. (1973) in The Enzymes (Boyer, P.D., Ed.),
3rd ed. Vol. 9, pp.1-48, Academic Press, New York.

Schwab, D. A., and Wilson, J.E. (1989) Proc. Natl. Acad.
Sci. USA 86, 2563-2567.

Thelen, A. P., and Wilson, J.E. (1991) Arch. Biochem.
Biophys. 286, 645-651.

Schwab, D. A., and Wilson, J.E. (1991) Arch. Biochem.
Biophys. 185, 365-370.

White, T. K., and Wilson, J.E. (1989) Arch. Biochem.
Biophys. 274, 375-393.

Griffin, L.D., Gelb, B.D., Wheeler, D.A., Davison, D.,
Adams, V., and McCabe, E.R.B. (1991) Genomics 11,
1014-1024.

Printz, R.L., Koch, S., Potter, L.R., O'Doherty, R.M.,
Tiesinga, J.J., Moritz, S., and Granner, D.K. (1993) J.
Biol. Chem. 268, 5209-5219.

Kogure, K., Shinohara, Y., and Terada, H. (1993) J. Biol.
Chem. 268, 8422-8424.

15. Mochizuki, Y. (1981) Comp. Biochem. Physiol. 703, 745-751.

16. Ureta, T., Reichberg, S.B., Radojkovic, J., and Slebe,

J.C. (1973) Comp. Biochem. Physiol. 458, 445-461.

24

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

25

Katzen, R.M., Soderman, D.D., and Wiley, C.E. (1970) J.
Biol. Chem. 245, 4081-4096.

Luech, J.D., and Fromm, H.J. (1974) J. Biol. Chem. 249,
1341-1347.

Xie, G., and Wilson, J.E. (1988) Arch. Biochem. BiOphys.
267:803-810.

'Felgner, P.L., and Wilson, J.E. (1977) Arch. Biochem.

Biophys. 182:282-294.

Kabir, F., and‘Wilson, J.E. (1993) Arch. Biochem. Biophys.
300, 641-650.

BeltrandelRio, H., and Wilson, J.E. (1991) Arch. Biochem.
Biophys. 286, 183-194.

Preller, A., and Wilson, J.E. (1992) Arch. Biochem.
Biophys. 294, 482-492.

Katzen, H.M., and Schimke, R.T. (1965) Proc. Natl. Acad.
Sci. U.S.A. 54, 1218-1225.

Printz, R.L., Magnuson, M.A., and Granner, D.K., Annu.
Rev. Nutr. 13, 463-496.

Miwa, I., Mitsuyama, S., Toyoda, Y., Nonogaki, T., Aoki,
s., and Okuda, J. (1990) Biochem. Int. 22, 759-767.

Purich, D.L., and Fromm, H.J. (1971) J. Biol. Chem. 246,
3456-3463.

Ellison, W.R., Luech, J.D., and Fromm, H.J. (1974)
Biochem. Biophys. Res. Commun. 57, 1214-1220.

Gregoriou, M., Trayer, I.P., and Cornish-Bowden (1983)
Eur. J. Biochem. 134, 283-288.

Ganson, N.J., and Fromm, H.J. (1985) J. Biol. Chem. 260,
12099-12105.

Copley, M., and Fromm, H.J. (1967) Biochemistry 6,
3503-3509.

Casazza, J.P., and Fromm, H.J. (1976) Arch. Biochem.
Biophys. 177, 480-487.

Fromm, H.J., and Zewe, V. (1962) J. Biol. Chem. 237,
1661-1667.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

26

Ellison, W.R., Lueck, J.D., and Fromm, H.J. (1975) J.
Biol. Chem. 250, 1864-1871.

Wilson, J.E. (1979) Arch. Biochem. Biophys. 196, 79-87.

Segel, I.H. (1975) Biochemical Calculations, 2nd ed., John
Wiley, New York.

Wilson, J.E. (1973) Arch. Biochem. Biophys. 159, 543-549.

Tsai, H.J., and Wilson, J.E. (1995) Arch. Biochem.
Biophys. 316, 206-214.

Radojkovic, J., and Ureta, T. (1987) Biochem. J. 242,
895-903.

Sols, A., and Crane, R.K. (1954) J. Biol. Chem., 210,
581-595.

Crane, R.K., and Sols, A. (1954) J. Biol. Chem., 210,
597-606.

Sols, A. (1976) in Reflections in Biochemistry (Kornberg,
A., Horecker, B.L., Cornudella, L., and.Gro, J., eds), pp.
199-206, Pergamon, Oxford.

Ureta, T., Lazo, P.A., and Sols, A. (1985) Arch. Biochem.
Biophys. 239, 315-319.

Anderson, C.M., Stenkamp, R.E., McDonald, R.C., and
Steitz, T.A. (1978) J. Mol. Biol. 123, 207-219.

Anderson, C.M., Zucker, F.H., and Steitz, T.A. (1979)
Science, 204, 375-380.

Bennett, W., and Steitz, T.A. (1980) J. Mol. Biol. 140,
211-230.

Harrison, R.W. (1985) Ph.D. thesis, Yale University, New
Haven, CT.

Schwab, D.A., and Wilson, J.E. (1989) Proc. Natl. Acad.
Sci. U.S.A. 86, 2563-2567.

Arora, K.K., Filburn, C.R., and Pedersen, P.L. (1993) J.
Biol. Chem. 268, 18259-18266.

Robert, S.C., Harrison, R.W., Bell, G.I., Pilkis, S.J.,
and Weber, I.T. (1994) Diabetes 43, 784-791.

Pilkis, S.J., Weber, I.T., Harrison, R.W., and Bell, G.I.
(1994) J. Biol. Chem. 269, 21925-21928.

52.

530

54.

550

56.

57.

58.

590

60.

61.

62.

63.

64.

65.

27

Zhong, L., Harrison, R.W., Weber, I.T., and Pilkis, S.J.
(1995) J. Biol. Chem. 270, 9939-9946.

Arora, K.K., Filburn, C.R., and Pedersen, P.L. (1991) J.
Biol. Chem. 266, 5359-5362.

Baijal, M., and Wilson, J.E. (1992) Arch. Biochem.
Biophys. 289, 271-278.

Chen, L., Zeng, C., and Fromm, H.J. (1993) FASEB J. 7,
A1176.

White, T. K., and Wilson, J.E. (1987) Arch. Biochem.
Biophys. 259, 402-411.

Yanagawa, H.A. (1978) Insect. Biochem. 8, 293-305.

Shoemaker, C.B., Reynolds, S.R., Wei, G., Tielens, A.G.M.,
and Harn, D.A. (1995) Exp. Parasitol. 80, 36-45.

Mochizuki, Y., and Hori, S.H. (1977) J. Biochem. 81,
1849-1855.

Chou, A. C., and Wilson, J.E. (1974) Arch. Biochem.
Biophys. 163, 191-199.

Hutny, J., and Wilson, J.E. (1990) Arch. Biochem. Biophys.
283, 173-183.

Solheim, L.P., and Fromm, H.J. (1981) Arch. Biochem.
Biophys. 211, 92-99.

Jarori, G.K., Iyer, S.B., Kasturi, S.R., and.Kenkare, U.W.
(1990) Eur. J. Biochem. 188, 9-14.

Tsai, H.J., and Wilson, J.E. (1996) Arch. Biochem.
Biophys. 329, 17-23.

Tsai, H.J., and Wilson, J.E. (1996) Arch. Biochem.
Biophys. (in press).

Chapter II
Functional Organization of Mammalian Hexokinases:
Characterization of Chimeric Hexokinases Constructed from
the N- and C- Terminal Domains of the Rat Type I and Type II

Isozymes

Archives of Biochemistry and Biophysics

Vol. 316, No. 1, Jan. 10, pp. 206-214, 1995

28

ABSTRACT

Chimeric hexokinases consisting of either the N-terminal
half of Type I hexokinase fused with the C-terminal half of
the Type II isozyme (NICII) or the inverse pair (NIICI), along
with the parental isozymes, were expressed.in COS-1 cells. The
thermal stability of the chimeras was intermediate between
that of the highly labile Type II isozyme and the relatively
stable Type I hexokinase. In their Kfis for substrates, Glc and
ATP, the chimeric enzymes were similar to the parental isozyme
from which the C-terminal half was derived. Although the Type
I and Type II isozymes were similar in their sensitivity to
inhibition (competitive vs. ATP) by the Glc-6-P analogs, 1,5-
anhydroglucitol 6-phosphate (AnGlc-6-P) and Glc-1,6-
bisphosphate, the chimeric enzymes differed markedly, with the
NIICI chimera being much more sensitive and the NICII chimera
much less sensitive than either parental form to these
inhibitors. In contrast, the response of the chimeras to P“
either as an antagonist of inhibition by AnGlc-6-P or, at
higher concentrations, as an inhibitor, was correlated with
the origin of the N-terminal domain. The results are
consistent with the view that catalytic function.is.associated
with the C-terminal domain of the Type I isozyme, with
regulatory function - inhibition by G1c-6-P and its analogs
and antagonism of this inhibition by Pi- being mediated by the

N-terminal domain.

29

30

Mammalian tissues contain three isozymes of hexokinase
(ATP: D-hexose 6-phosphotransferase, EC 2.7.1.1), generally
designated as Types I-III, which exhibit marked similarity in
some of their properties (reviewed in ref. 1). These include
existence as monomeric enzymes with melecular weights of
approximately 100,000, and sensitivity to inhibition by
physiological levels of the product, Glc-6-PK The
availability of amino acid sequences for the Type I, II and
III isozymes, deduced from the nucleotide sequence of cDNAs
cloned from various sources (2-8) , has revealed extensive
sequence similarity among the mammalian isozymes, confirming
that they are all closely related members of a family which
also includes other kinases (9). Moreover, the sequences of
these isozymes exhibit internal repetition, i.e., the.sequence
of the N-terminal half is similar to that of the C-terminal
half, and further, these sequences show similarity to that of
various 50 kDa hexokinases that exist in organisms such as
yeast (1,10,11). These results are consistent with the
suggestion (12-15) that the 100 kDa mammalian enzymes evolved
by duplication and fusion of a gene encoding an ancestral 50
kDa hexokinase.

The extensive similarity in amino acid sequences of the
N- and C-terminal halves of the 100 kDa mammalian hexokinases
clearly implies a corresponding similarity in secondary and
tertiary structure (16). Moreover; a number of residues which,

based on earlier work with the homologous yeast hexokinase

31

(17), are known to be of«catalytic importance are conserved in
both N- and C-terminal halves of all mammalian hexokinases
that.have been sequenced thus far (1-8). Despite this striking
similarity in sequence and, presumably, structure, there is
clear evidence that, at least for the Type I isozyme, the N-
and C-terminal halves of the enzyme are functionally distinct.
Thus, White and Wilson (15) demonstrated that the N- and C-
terminal halves of the rat Type I isozyme could be isolated
after limited proteolysis, with complete retention of
catalytic activity in the C-terminal half while the isolated
N-terminal half was devoid of detectable activity. Consistent
with this, site-directed mutation of catalytically important
residues in the C-terminal half of the rat (18) or murine
(19,20) Type I isozyme result in drastic reductions of
catalytic function, while mutation of analogous residues in
the N-terminal halves have no effect on catalysis (18,20).
Finally, the C-terminal halves of Type I hexokinase, expressed
in.E. coli using the cloned cDNAs for the murine (20) or human
(21,22) isozymes, exhibit catalytic activity while the
corresponding expressed N-terminal half does not (20).

While catalytic function has unequivocally been
associated with the C-terminal half of the Type I isozyme,
assignment of regulatory function to a particular structural
region has been much less straightforward and remains
controversial. This has been extensively reviewed elsewhere

(1) but basic elements of the debate are summarized here.

32

White and Wilson (23,24) demonstrated the presence of a
Glc-6-P binding site in the N-terminal half; the affinity (Kd
z 10 uM) of this site for Glc-6-P was virtually identical to
the kinetically determined Ki, and the binding of various
hexose 6-phosphates to this site, with resulting effect on
conformation of the N-terminal half, was correlated with their
inhibitory effectiveness. Hutny and Wilson (25) showed that
binding of Glc-6-P preferentially protected sulfhydryl groups
in the N-terminal half of the enzyme from reaction with a
sulfhydryl-selective reagent, again with effective
concentrations in the low pH range. Previous binding studies
(26-28) had shown a single site (per 100 kDa molecule) for
binding of Glc-6-P with aJK,in the low uM range. Thus, the N-
terminal site was proposed as the regulatory site. According
to this view, inhibition is considered to result from
conformational changes, induced by the binding of Glc-6-P to
the N-terminal half, which preClude binding of ATP at the
catalytic site in the C-terminal half (24,25).

The situation was complicated by the finding that the
isolated (by limited proteolysis) ( 15) or expressed (from
cloned cDNAs) (20-22) C-terminal half of the enzyme retained
sensitivity to inhibition by Glc-6-P, leading to suggestions
(20,22) that the functional regulatory site is in the C-
terminal half, with the role of the Glc-6-P binding site in
the .N-terminal half unspecified. However, based on this

suggestion, one would expect two binding sites (per 100 kDa

33

molecule) for Glc-6-P, which conflicts with the observed
stoichiometry (26-28). An alternative view (24), consistent
with the stoichiometry, is that the G1c-6-P binding site in
the C-terminal half of the molecule is latent in the intact
100 kDa enzyme, with the N-terminal site serving regulatory
function.

Whether the functional organization proposed (15, 23-25)
for the Type I isozyme, i.e., catalytic C-terminal half and a
regulatory' N-terminal half, is parallelled in 'the other
isozymes of mammalian hexokinase remains to be determined. In
the present study, we take one approach toward addressing this
question. We have prepared constructs that permit expression
of chimeric hexokinase molecules, comprised of the N-terminal
half of Type I hexokinase and the C-terminal half of the Type
II isozyme, and vice versa. We ask whether regulatory
characteristics of the chimeric enzymes can be correlated with
those of the parental enzyme from which.the.N-terminal half of
the chimera was derived. For the Type I and Type II isozymes,
a.major distinguishing characteristic is their response to PP
With Type I hexokinase, low concentrations of Pi antagonize the
binding of Glc-6-P and the inhibition resulting from binding
of this ligand (26,27,29). At higher concentrations, Piitself
becomes inhibitory, competitive with ATP (27). As noted by
Ellison et al. (27), this bimodal action of 1% implies the
existence of two distinct sites for binding of this ligand. It

has been suggested that the high affinity site is located in

34

the N-terminal half of the molecule (24,25), with competition
between Piand the 6-phosphate group of Glc-6-P for a common
anion binding site accounting for the mutually exclusive
binding of these ligands (26,27) and thereby, reversal of the
inhibition by Glc-6-P (26,27,29). The second site, with lower
affinity and accounting for the inhibition of Type I
hexokinase at elevated concentrations of P“ is suggested to
be located in the C-terminal half (24,25). In contrast to this
dual action on the Type I isozyme, Pi does not antagonize
inhibition of Type II hexokinase by Glc-6-P, but is solely
inhibitory, again competitive with ATP, at concentrations

throughout the mM range (30).
‘MATERIALS AND METHODS

Materials. Restriction enzymes, T4 DNA ligase, and Glc-6-
P dehydrogenase were products of Boehringer Mannheim
(Indianapolis, IN). BCA.Protein Assay'Reagent and.bovine serum
albumin standard were purchased from Pierce Chemical Co.
(Rockford, IL) and all other biochemicals from Sigma Chemical
Co. (St. Louis, MO). Isoelectric focusing was performed on pH
3-10 gels obtained from Serva (Paramus, NJ) and cellulose
acetate electrophoresis on Titan III plates from Helena
Laboratories (Beaumont, TX).

geeetruction of vectors and expression of chimeric

hexokinases in COS-1 cells. Procedures for restriction

35
digestion, ligation of DNA and transformation of E. coli (DH5-
a) were essentially as described by Sambrook et al. (31).

The previously described (3) 3.6 kb full length cDNA for
rat Type I (HKI 1.4-7) was cloned into the Eco RI site of
pUC18. An Nco I site (Fig. 1) is conveniently located near the
center of the coding region (cleavage at nucleotide 1452),
while a Pst I site (cleavage at nucleotide 3156) is located
approximately 300 bp past the termination codon. Thus,
digestion with Eco RI and Nco I gave a fragment of about 1.4
kb containing the coding sequence for the N-terminal half Of
the molecule, and digestion with Nco I and Pst I gave a 1.7 kb
fragment containing the coding sequence for the therminal
half of the enzyme.

As previously described (6), the nucleotide sequence
encoding the rat Type II isozyme was contained in two
overlapping clones, 12-1.3C coding for the N-terminal half of
the molecule and RG2B coding for the C-terminal half. A full
length cDNA for the rat Type II isozyme was generously
provided by Dr. Annette Thelen. It had been constructed by
ligation of complementary fragments from 12-1.3C and RG2B. In
addition, a thymidine corresponding to nucleotide 1559 in the
full length cDNA had been mutated to cytosine to yield an Nco
I site at a position equivalent to that in the cDNA encoding
the Type I isozyme (Fig. 1). Thus, digestion with Eco RI-Nco
I gave a fragment of about 1.5 kb containing the coding

sequence for the N-terminal half of Type II hexokinase, while

36

Type I 450 Gly Lys Gly Ala Ala Met Val Thr Ala
Type I 1439 GGC AAG GGG GCC GQC ATG GTG ACG GCA
Type II 1545 GGC AAG GGG GCT GCT ATG GTG ACG GCG
Type II 450 Gly Lys Gly Ala Ala Met Val Thr Ala

Figure 1. Nucleotide and deduced amino acid sequences
surrounding the centrally located Nco I sites in cDNAs for rat
Type I and Type II hexokinases. Nco I cleaves after the
cytosine:residue that is underlined.in the nucleotide sequence
for the Type I isozyme. The thymidine residue that was mutated
to cytosine to generate an Nco I site in the Type II cDNA is
shown in bold.

37
digestion with.Nco I-Pst I gave a 1.5 kb fragment encoding the
C-terminal half of the Type II isozyme.

At both. the nucleotide and amino acid levels, the
sequences of the Type I and II isozymes were virtually
identical in the region surrounding the central (and sole) Nco
I site (Fig. 1). Chimeric cDNAs 'were constructed. after
digestion of plasmids containing cDNAs for the Type I and Type
II isozymes with Nco I-Pst I, excising the regions
corresponding to the C-terminal halves of the parental
isozymes. The Nco I-Pst I fragments were gel isolated, and
religated back into the complementary larger fragment (vector
plus cDNA region encoding N-terminal half of the molecule)
from the Nco I-Pst I digestions. The chimeric cDNAs were
cloned in.pUC18, then subcloned into the vector pSVT7 (32) for
expression in COS-1 cells.

COS-1 cells were cultured as previously described (18),
and transfected (10 ug plasmid DNA per 100 mm culture dish)
following the procedure of DeWitt et al. (33) with addition.of
a 90 sec treatment with dimethylsulfoxide at the end of
transfection (34) . Control cultures were untransfected or were
transfected with the pSVT7 vector alone (sham transfected);
both gave equivalent results. Cells were harvested 2-3 days
after transfection, suspended in 1 ml (per 100 mm dish) of
ice-cold BTGE buffer (50 mM Bicine, 10 mM thioglycerol, 10 mM
Glc, 0.5 mM EDTA, pH 8.2) and sonicated for 10 sec. Sonicates

were centrifuged at 800 x g for 10 min at 4° C. Hexokinase

38
activity in the supernatants was determined immediately, and
supernatants stored at -80° C for further use. Where necessary
(e.g. , for isoelectric focusing or cellulose acetate
electrophoresis) , samples were concentrated using a Centricon-
30 device (Amicon Corp., Beverly, MA). A

Preparation of brain extract containing Tym I
hexokiease. After decapitation of a Sprague-Dawley rat (either
sex, approx. 200 9) under CO2 anesthesia, the brain was
rapidly removed and homogenized in BTGE buffer (3 ml per 9
brain) using a Teflon-glass homogenizer (A.H. Thomas Co.,
Philadelphia, PA). The homogenate was centrifuged at 4° C and
3000 x g for 5 min, and the fluffy white layer (mostly myelin)
removed. The remainder of the homogenate (with pellet
resuspended in the supernatant) was made 2 mM in Glc-6-P and
incubated for 30 min at room temperature to promote release of
the mitochondrially bound hexokinase (35) . After
centrifugation at 164,000 x g for 1 hr, the solubilized
hexokinase was recovered in the supernatant.

Qetermieation of hexokinase activity and ereteig.
Hexokinase activity was assayed as described previously (35) .
Determination of Kms for Glc and ATP were done under the same
conditions, except that the specified substrate was varied
over an appropriate concentration range. Prior to
determination of the K“I values and heat stability, cell
extracts were chromatographed on spin columns of Sephadex G-25

(fine) equilibrated with HET buffer (50 mM Hepes, 0.5 mM EDTA,

39

10 mM thioglycerol, pH 7.5); absence of Glc in the eluted
enzyme was confirmed by enzymatic assay. Inhibition by the
Glc-6-P analog, AnGlc-6-P, was determined as specified by
Baijal and Wilson (18); AnGlc-6-P is not a substrate for Glc-
6-P dehydrogenase and thus the convenience of the standard
coupled spectrophotometric assay could be retained. Inhibition
by Glc-1,6-bisphosphate was studied in a similar manner,
maintaining the concentration of Mg“+ at 0.5 mM excess over
the varied ATP concentration. All kinetic data were analyzed
using the EZ Fit program of Perrella (36).

Protein was determined with the BCA Protein Assay system
from Pierce Chemical Co., using bovine serum albumin as
standard. Samples were pretreated with excess iodoacetamide
(37) to avoid interference by thioglycerol present in the
BTGE buffer.

Electrophoresis and isoelectric focueipg. SDS acrylamide
gel electrophoresis and immunoblotting procedures were as
described by Smith and Wilson (38). Electrophoresis under
nondenaturing conditions was performed on Titan III cellulose
acetate plates equilibrated with TBPE buffer (180 mM Tris
base, 100 mM bOric acid, 20 mM NaZHPO“ 2 mM EDTA) . Extracts of
transfected cells were concentrated to approximately 2 units
hexokinase/ml; extracts from sham transfected cells were
concentrated to a comparable extent on a volume basis but, of
course, the hexokinase activity was much lower. Samples were

loaded using a Zip Zone sample applicator (Helena

40

Laboratories, Beaumont, TX); the exact amount of sample loaded
cannot be controlled with this applicator, but was estimated
to be approximately 1 pl. Electrophoresis in TBPE buffer was
performed at 175 volts for 30 min.at 4° C. Hexokinase activity
was detected.by covering the plate with.a warm (approx. 45° C)
solution containing 1% agar, 45 mM TrisCl (pH 9.5), 7.4 mM
MgC12, 3.7 mM Glc, 0.02 mg/ml phenazine methosulfate, 0.25
mg/ml nitro blue tetrazolium, 0.7 mg/ml NADP, and 3.2 units/ml
Glc-6-P dehydrogenase; the last four components were added
immediately before use. When the agar had solidified, the
plate was incubated in the dark at room temperature until the
bands of activity were evident; staining was stopped by
removing the agar layer and placing the plates in 10% acetic
acid.

Isoelectric focusing was performed on pH 3-10 gradient
gels, with an anode buffer containing 3.3 g aspartic acid and
3.7 g glutamic acid per 1 and a cathode buffer containing 4 g
arginine and 4 g lysine per 1. Samples (5 ul) contained
approximately 0.01 units of activity. Focusing was conducted
for 2 hrs with gradual increase of power: 1 watt for 30 min,
5 watts for 60 min, then 10 watts for 30 min. Hexokinase
activity was detected by incubation of the focusing gels in a
solution as used for staining of cellulose acetate plates (see
above) except that agar was deleted and 1% (v/v) Triton X-100
was added. Staining was stopped by placing the gel in 10%

acetic acid.

41
gelculation of isoelectric pointe. Isoelectric points of
the Type I and Type II isozymes, and their N- and C-terminal
halves, were calculated from the amino acid sequences (3,6)
using the Isoelectric routine included in the GCG Sequence
Analysis Software Package (Genetics Computer Group, Inc.,

Madison, WI)
RESULTS

We will refer to chimeric hexokinases by symbolism that
identifies the origin of the N- and C-terminal halves. Thus,
NICII refers to the chimeric enzyme comprised of the N-
terminal half (residues 1-454) of the Type I isozyme and the
C-terminal half (residues 455-917) of Type II hexokinase, and
NIICI represents the chimera comprised of residues 1-454 from
Type II hexokinase and residues 455-918 from Type I
hexokinase.

Zegification of chimeric cDNA conetructe. The chimeric
constructs, along with the parental cDNAs, were examined by
restriction analysis (Fig. 2). Digestion with Eco RI and Nco
I gave fragments of the size expected for regions encoding the
N-terminal halves of the Type I and II isozymes, 1.4 kb and
1.5 kb, respectively. Digestion with Nco I and Pst I gave
fragments of 1.7 kb and 1.5 kb, corresponding to the coding
regions for the C-terminal halves of the Type I and Type II

hexokinase, respectively. In all cases, the chimeric cDNAs

42

 

N-Terminal C-Terminal
(Eco RI/Nco I) (Nco I/Pst I)

Figure 2. Restriction analysis of parental cDNAs and chimeric
constructs. Fragments of interest are indicated at left and
right of figure. Lanes 1-4, products of digestion with Eco RI
and Nco I, which release 1.4 kb and 1.5 kb fragments encoding
the N-terminal halves of the Type I and Type II isozymes,
respectively. Lane 1, plasmid containing cDNA for Type I
isozyme; Lane 2, plasmid containing cDNA for Type II isozyme;
Lane 3, plasmid containing cDNA for NICII chimera; Lane 4,
plasmid containing cDNA for NIICI chimera. Lanes 5-8
correspond to Lanes 1-4, but with digestion by Nco I and Pst
I to release 1.7 kb and 1.5 kb fragments encoding the C-
terminal halves of the Type I and Type II isozymes,
‘ respectively.

43
gave fragments of the size.expected from the parental origins
of the N- and C-terminal halves.

Expgession of chimeric hexokinases in COS-1 celle.
Extracts of COS-1 cells that had been transfected with pSVT7
constructs containing cDNAs for the Type I or Type II isozymes
or for chimeric hexokinases all showed hexokinase activities
that were well above that seen in extracts from sham
transfected controls (Table I) . Although the variation between
experiments was considerable, the lspecific activities in
extracts from cells transfected with cDNA for Type II
hexokinase or for the NIICI chimera were consistently somewhat
higher than those for cells transfected with cDNA for Type I
hexokinase or the NICII chimera; by two-way ANOVA and
Bonferroni’s t-test, the Type II and NIICI values differed
from the others with p<0.01. This could not be attributed to
expression of higher levels.of Type II hexokinase or the NIICI
chimera since all forms were expressed.at similar levels based
on intensity of staining with Coomassie Blue or on Western
blots (see below). Thus, it would appear that the presence of
the N-terminal half of the Type II isozyme somehow leads to
increased specific activity. One obvious possibility is that
the N-terminal half of the Type II isozyme, unlike that of
Type I hexokinase (15,18, 19-22) , possesses intrinsic catalytic
activity. This is currently being investigated but, if true,
both halves must be quite similar in kinetic properties (39)

since no departure from simple Michaelis-Menten behavior for

44

Table I. Hexokinase Activity in Extracts of COS-1 Cells
Transfected.with cDNAs for Type I or Type II Hexokinase or for

Chimeric Hexokinases.

cDNA Used for
We

Sham Control
Type I

Type II
NICII

NIICI

0.08

i

i

H-

H-

H-

Hexokinase Activity‘

(unitszmg protein)

 

‘Mean 1 SD for 11 different transfections.

45
the Type II isozyme has been observed in.previous studies (40)
or in the present work (results presented below). The
observation that the NIICI chimera had Khs for Glc and ATP
(see below) that were intermediate between those of the
parental Type I and Type II isozymes is consistent with this
possibility (39).

All extracts from transfected cells contained a prominent
component with apparent molecular weight of approximately
100,000 (Fig. 3). No intense band was seen in this region in
control extracts from untransfected or sham transfected COS-1
cells, and thus these components can reasonably be assumed to
correspond to the expressed hexokinases (supported by
immunoblotting results, presented below). The mobility of the
Type I isozyme expressed in COS-1 cells was indistinguishable
from that of the Type I isozyme isolated from rat brain (35).
Although the Type I and Type II isozymes are virtually
identical in size - 918 residues with calculated.M,of 102,317
for Type I (3) and 917 residues and calculated Mr of 102,555
for Type II (6) - the expressed Type II isozyme consistently
migrated slightly faster than did the Type I isozyme; this
same result has been seen with the Type II isozyme isolated
from. rat skeletal muscle (A.P. Thelen and J.E. Wilson,
unpublished observation). The slight difference in mobility of
these isozymes on SDS gels presumably results from charge
differences which provide the basis for resolution of the

isozymes by electrophoresis under nondenaturing conditions

46

 

123456

 

 

 

 

Figure 3. SDS gel electrophoreth: analysis of COS-1 cell
extracts. Lane 1, Type I hexokinase purified from rat brain
(35); Lane 2, extract from sham transfected COS-1 cells; Lanes
3-6, extracts from COS—1 cells transfected with cDNAs for Type
I, Type II, NICII, and NIICI hexokinases, respectively.

47
(41) or ion exchange chromatography (42) and which, despite
the presence of SDS, influence the electrophoretic migration
on SDS gels. The chimeric hexokinases (Fig. 3, Lanes 5 and 6)
migrated. to positions intermediate between those of the
parental isozymes.

The identity of the chimeric hexokinases was confirmed by
Western blots (Fig. 4) probed with monoclonal antibodies 21
and 5A, previously shown (43) to recognize segmental epitopes
located in the N- and C-terminal halves, respectively, of the
Type I isozyme. With both antibodies, extracts from COS-1
cells transfected with the Type I cDNA showed an
immunoreactive band migrating identically to authentic Type I
hexokinase purified from rat brain (35), while extracts from
sham transfected cells showed no such band. Extracts from
cells transfected with the cDNA for Type II hexokinase showed
no detectable reactivity with monoclonal antibody 21; with
prolonged staining, faint reactivity with monoclonal antibody
5A could be detected but this was clearly much less intense
than with the Type I isozyme. The chimeric hexokinases (Fig.
4, Lanes 5 and 6) displayed the expected immunoreactivity.
Thus, NICII (Lane 5) was intensely stained with monoclonal
antibody 21 but only weakly detected with monoclonal antibody
5A; conversely, NIICI (Lane 6) was intensely reactive with
monoclonal antibody 5A but showed no detectable reactivity
with monoclonal antibody 21. The immunoreactive bands seen

with extracts from cells transfected with the chimeric cDNAs

48

 

Figure 4. Western blot analysis of COS-1 cell extracts. Lane
1, Type I hexokinase purified from rat brain (35); Lane 2,
sham tranfected COS-1 cells; Lanes 3-6, extracts from COS-1
cells transfected with cDNAs for Type I, Type II, NICII, and
NIICI hexokinases, respectively. Top panel, blot probed with
monoclonal antibody 21, which recognizes an epitope located in
the N-terminal half of Type I hexokinase (43). Bottom panel,
blot probed with monoclonal antibody 5A, which recognizes an
epitope located in the C-terminal half of Type I hexokinase
(43). Weak reactivity of antibody 5A.with the Type II isozyme
is detectable; this antibody also recognizes another component
of lower molecular weight, present in extracts from both sham
transfected (Lane 2) and transfected (Lanes 3-6) cells.

49

were, as on the Coomassie Blue stained gel (Fig. 3), located
at positions intermediate between the parental Type I and.Type
II isozymes. A minor immunoreactive species, with M,.of
approximately 90,000, was detected with monoclonal antibody 21
in extracts from cells transfected with the Type I and NICII
cDNAs; this is likely a result of proteolysis by endogenous
proteases that cleave within the N-terminal domain, as
previously shown to occur in brain homogenates (35).

Isoelectric focusing (Fig. 5) and electrophoresis under
nondenaturing conditions (Fig. 6) revealed a single band of
hexokinase activity in extracts from COS-1 cells transfected
with either the parental or chimeric cDNAs; endogenous
activity in extracts from sham transfected cells was too low
to be detected in these procedures. The isoelectric point and
electrophoretic mobility of the expressed Type I hexokinase
was identical to that of authentic Type I hexokinase from rat
brain, while the Type II isozyme exhibited a more acidic p1
and greater electrophoretic mobility consistent with its
previously described electrophoretic and ion exchange
chromatographic behavior (41,42). The chimeric hexokinases
exhibited p15 and electrophoretic mobilities intermediate
between those of the parental isozymes. It is interesting to
note that the charge properties of both the chimeras and
parental enzymes appear to be dominated by the N-terminal
half, i.e., NICII was more similar to the Type I istyme,

while NIICI was more similar to the Type II isozyme. This is

 

50

 

 

 

Figure 5. Analysis of COS-1 cell extracts by isoelectric
focusing. Lane 1, Type I hexokinase in extract from rat brain;
Lane 2, extract from sham transfected COS-1 cells; Lanes 3-6,
extracts from cells transfected with cDNAs for Type I, Type
II, NICII, and NIICI hexokinases, respectively. After
focusing, the gel was stained for hexokinase activity.

51

 

 

 

Figure 6. Analysis of COS-1 cell extracts by electrophoresis
under nondenaturing conditions. Lanes 1 and 7, Type I
hexokinase in extract from rat brain; Lane 2, extract from
sham transfected COS-1 cells; Lanes 3-6, extracts from cells
transfected with cDNAs for Type I, Type II, NICII, and NIICI
hexokinases, respectively. After electrophoresis, the
cellulose acetate plate was stained for hexokinase activity.

52
consistent with p15 calculated from the amino acid sequences,
which are 6.4 and 5.7 for the N-terminal halves of the Type I
land Type II isozymes, respectively, while the C-terminal
halves are considerably less different in calculated pIs,
being 6.7 and 6.3 for the Type I and Type II isozymes,
respectively.

Heat stability of Type I, Type II, and chimeric
hexokinases. Grossbard and Schimke (40) first noted the marked

 

difference in heat stability of the isozymes of mammalian
hexokinase, with the Type II isozyme being much more labile
than the Type I isozyme. This was also seen with the Type I
and Type II isozymes expressed in COS-1 cells (Fig. 7). Both
chimeric enzymes exhibited similar thermal stability that was
intermediate between that of the parental isozymes. It is
notable that the chimeric enzymes were not more labile than
the Type II isozyme. Thermal stability is reasonably taken as
at least a gross measure of overall structural integrity; by
this criterion, creation of the chimeric proteins has not led
to marked structural destabilization. Structural integrity is
also implied by the preservation of parental-like kinetic
properties in the chimeric enzymes.

Kinetic and reglatory progrties of Tyg I . TZE II , and
chimeric hexokinases. The Kms of the Type I and Type II
isozymes, and their chimeras, for Glc and ATP are shown in
Table II. The results are quite similar to those reported by

Grossbard and Schimke (40) and others (14) for the rat Type I

53

 

5
'1
II
II

75- \.\ . I

\V

 

> A

'8‘ 50 _

<1 .3 _

82 . ; . NIICI .
25 - H ‘ NICII

0

 

 

 

o I l l I
O 5 10 15 20

Time (min)

Figure 7. Heat stability of Type I, Type II, and chimeric
hexokinases. Hexokinase activity in extracts of COS-1 cells
transfected with cDNAs for Type I (Cl) , Type II (0 ) , NICII

(A) , and NIICI (V) hexokinases was determined as a function
of time at 40° C. Values shown are the average from two
experiments (each with extracts from a different transfection)
which gave similar results; average deviation between
experiments was i 4% from the values shown.

54

Table II. Kinetic Properties of Type I and Type II Hexokinase
and the Chimeric Hexokinases.

Km for Ki for
Ieeeyme Glucose(pM)‘ ATP (mM)‘ Angie-g-P (gu)°
Type I 65 i 4 0.52 i 0.04 17 i 1
Type II 142 i 1 0.77 i 0.07 22 i 4
NICII 140 I 17 1.01 i 0.15 72 i 20
NIICI 82 i 13 0.64 i 0.06 6 i 1

 

‘Mean 1 SD for 3 determinations, each with an extract from a
different transfection.
bMean 1 SD for 4 determinations, each with an extract from a
different transfection.

55
and Type II isozymes, and indicate somewhat higher apparent
affinity (lower Kh) of the Type I isozyme for both substrates.
Kinetically, the chimeras resembled the parental isozyme from
which the C-terminal half was derived, i.e., NIICI had a lower
Km for both substrates than did NICII.

As previously found for Glc-6-P itself (14,40) as well as
its analog, AnGlc-6-P (44,45), the Type:I and Type II isozymes
were quite similar in their sensitivity to inhibition by
AnGlc-6-P (Table II). In contrast to the parental isozymes,
the chimeric enzymes differed markedly in their sensitivity to
this inhibitor, with NICII having aim approximately 10-fold
greater than that seen with NIICI. For all forms, inhibition
by AnGlc-6-P was competitive vs. ATP, as found in earlier
studies with the parental isozymes (44,45).

In contrast to the similarity in sensitivity of the Type
I and Type II isozymes to inhibition by AnGlc-6-P (44, and
Table II), Rose et al. (44) had found that the Type II isozyme
was considerably more sensitive to inhibition by Glc-1,6-
bisphosphate. This was confirmed in the present study (Fig.
8). As in the case of inhibition by AnGlc-6-P, the chimeric
enzymes differed much more markedly than did the parental
isozymes in their sensitivity to inhibition by Glc-1,6-
bisphosphate, with the NIICI chimera again being much more
susceptible than was the NICII chimeric hexokinase (Fig. 8).
In agreement with Rose et a1. (44), inhibition was found to be

competitive vs. ATP (data not shown) ; Ki values determined (one

56

 

100 '3: ‘5 y T
T .3 .-. NICII
75 - i
a I
-.?.-. I
0 so - '
< ~ 11 .
°\° .L
. 25 - NIICI °

 

 

 

o I l J l
0 25 50 75 100

" Glc-1,6-bisP (uM)

Figure 8. Inhibition of Type I, Type II, and chimeric
hexokinases by Glc-1,6-bisphosphate. Activity was determined
as 'a function of increasing Glc-1,6-bisphosphate
concentration, with initial ATP concentration in each assay
equal to the Km value (Table II) for the respective
hexokinase. ([3), Type I; (0), Type II; (A), NICII; (V),
NIICI. Values shown are mean f SD from three experiments, each
with extracts from a different transfection; where no error
bars are seen, they are obscured by the data point symbol.

57
experiment) for the Type I, Type II, NICII, and NIICI
hexokinases were 40, 11, 143, and 4 pM, respectively.

Although.P,is an inhibitor, competitive vs. ATP, for both
Type:I and Type II hexokinase, the Type I isozyme was reported
to be much less sensitive to this inhibitor, with a K,of 35
mM (27), well above the K, of 2.7 mM found for the Type II
isozyme (30). We have confirmed this marked difference in
sensitivity to P, with the Type I and Type II isozymes
expressed in COS-1 cells (Fig. 9). The chimeric enzymes were
indistinguishable from the parental isozyme from.which the N-
terminal half was derived, i.e., NICII responded as the Type
I isozyme, and NIICI as the Type II isozyme (Fig. 9).

This same correlation with the origin of the N-terminal
half was seen in the response of the chimeric enzymes to P,as
an antagonist of inhibition by the Glc-6-P analog, AnGlc-6-P
(Fig. 10). The Type I isozyme, expressed in COS-1 cells,
responded as previously reported for the Type I isozyme from
brain (26,27,29), i.e., low concentrations of P,reversed the
inhibition while at higher concentrations, inhibition by P,
itself became evident.

The NICII chimera was virtually identical to the Type I
isozyme in this response. In contrast, P, did not reverse
inhibition of the Type II isozyme, either from rat skeletal
muscle (30) or expressed in COS-1 cells (Fig. 10), and the
NIICI chimeric hexokinase was indistinguishable from the Type

II isozyme in this lack of response to P,as an antagonist of

58

 

100 »-

I E‘-
': 'L ““\
u ‘-
75 x fifififififififi I
*- u\ ..... I
z: NICII
eeq \
.2. n
«I _L .....
2 50 - ‘-~-_‘—__5
~ 7
82 NIICI .L

 

 

 

Phosphate (mM)

Figure 9. Inhibition of Type I, Type II, and chimeric
hexokinases by P,. Activity was determined as a function of
increasing P, concentration, with initial ATP concentration in
each assay equal to the Km value (Table II) for the respective
hexokinase. (Cl), Type I; (0), Type II; (A), NICII; (V),
NIICI; where no error bars are seen, they are obscured by the
data point symbOl.

59

 

100 I

'ty

1V1

 

% Act

 

 

 

 

Phosphate (mM)

Figure 10. Effectiveness of P,as an antagonist of inhibition
by' AnGlc-6-P. .Activity' was. determined as a function of
increasing P,concentration, with initial ATP concentration in
each assay equal to the KIn value (Table II) for the respective
hexokinase, and AnGlc-6-P concentration sufficient to reduce
the activity to approximately 50% of that seen in the absence
of AnGlC-G-P- (ID). Type I; (0), Type II; (A). NICII; (V).
NIICI. Values shown are mean f SD from three experiments, each
with extracts from a different transfection; where no error
bars are seen, they are obscured by the data point symbol.

60

inhibition by AnGlc-6-P.
DISCUSSION

There are at least two possible interpretations for the
observation that ichimeric hexokinases can be expressed as
highly active species with retention of kinetic
characteristics.and.thermal stabilities that are.comparable to
the parental isozymes. One would be that the N- and C-terminal
halves of these molecules exist as quasi-independent domains.
If this were the case, the properties of a particular domain
would essentially be independent of the other, e.g., the
catalytic activity of the C-terminal domain of Type I
hexokinase (15) would be fully expressed in chimeric forms,
and its catalytic and regulatory properties would not depend
on the identity of the N-terminal domain with which it was
fused. The present work demonstrates that the latter
prediction is not fulfilled. The inhibition by P, and the
effectiveness of P, as an antagonist of inhibition by AnGlc-6-P
are clearly dependent on the origin of the N-terminal domain.

Previous studies provide additional evidence against the
view that the N- and C-terminal halves are essentially
independent. Thus, if these regions were tethered together by
a linking polypeptide sequence but with no significant
noncovalent interactions between the halves, one might

anticipate the linking polypeptide segment to be readily

61

susceptible to proteolysis, as is frequently the case with
segments linking functionally and structurally independent
domains (46). In fact, cleavage*within the segment linking the
N- and. C-terminal halves of Type. I Ihexokinase requires
perturbation of the structure with low concentrations of a
denaturing agent, guanidine hydrochloride, implying that the
linking segment is "buried" in the native structure as a
result of close spatial interactions between the N- and C-
terminal regions (15). Moreover, if the N- and C-terminal
regions were quasi-independent, then one would anticipate that
conformational changes induced by the binding of ligands to a
particular domain would largely be restricted to that domain.
However, there is abundant evidence that binding of various
ligands, including hexoses, hexose 6-phosphates, ATP, and P“
evokes conformational effects throughout the molecule,
affecting susceptibility to proteolytic attack or chemical
modification of sulfhydryl or arginyl residues, thermal
stability, and immunoreactivity with monoclonal antibodies
recognizing conformational epitopes (25,47-51).

Collectively, these observations provide a solid basis
for rejecting the view that the N- and C-terminal halves
represent functionally and structurally discrete,- quasi-
independent entities. On the contrary, they lead to the
conclusion that the N- and C-terminal halves of these isozymes
are in intimate contact, providing a structural basis for

functional interactions between these regions. Thus, an

62

alternative interpretation for the preservation of catalytic
function in the chimeric enzymes would be that interactions
between the N- and C-terminal domains have been maintained,
without distortion of structural features critical for
catalysis. In other words, it seems likely that interactions
between the N- and C-terminal halves are similar in the Type
I and Type II isozymes. In view of the extensive similarity
between the amino acid sequences of these isozymes (and also
that of the Type III isozyme) (1), which obviously must
include segments involved in interactions between the N- and
C-terminal domains, it is perhaps not surprising that this
would be the case.

As noted in the introduction to this paper, the location
of the functional allosteric site t0‘which.the inhibitory Glc-
6-P is bound remains in dispute (20,22,23-25). The present
work obviously does not resolve this controversy since there
is no clear correlation between the relative susceptibility to
inhibition by the Glc-6-P analogs, AnGlc-6-P and Glc-1,6-
bisphosphate, and the parental origin of either the N- or C-
terminal domain of the chimeric hexokinases (but see below).
The present study demonstrates that the affinity for these
inhibitors is markedly influenced by the identity of the N-
and C-terminal domains comprising the 100 kDa enzymes. It thus
seems likely that the apparent affinity for inhibitory hexose
phosphates is determined by complex interactions involving

both the N- and C-terminal domains, and depends on specific

63
aspects of the interdomain interactions that, despite an
overall similarity, vary somewhat in both the parental
isozymes as well as the chimeras.

Such ambiguity is not seen in the response to P, as an .
inhibitor and as an antagonist of inhibition by AnGlc-6-P. It
is clear that this is correlated with the origin of the N-
terminal domain. The results are fully consistent with the
proposal (24,25) that the N-terminal half of Type I hexokinase
is the location of an anion binding site for which P,and the
6-phosphate of Glc-6-P (or analogs) compete, accounting for
both the mutually exclusive binding of these ligands (26,27)
and the resulting antagonism of inhibition by the hexose 6-
phosphates (with binding of P,at the N-terminal site.not being
inhibitory), while the C-terminal half includes a site to
which P, binds with lower affinity and with resulting
inhibition.

Finally, wenote that the present study has revealed that
the chimeric hexokinases Idiffer significantly from the
parental isozymes in their response to product inhibition,
i.e., the ‘marked sensitivity’ of the ‘NIICI chimera, and
relative insensitivity of the NICII chimera, to inhibition by
Glc-6-P analogs (and presumably Glc-6-P itself). This offers
an intriguing opportunity for examining the allosteric
regulation of hexokinase activity in situ by transfection of
cells with DNA. encoding" hexokinase forms differing

substantially in their sensitivity to product inhibition.

10.

11.

12.

13.

14.

15.

16.

REFERENCES

Wilson,.J.E. (1994) Rev. Physiol. Biochem. Pharmacol., in
press.

Nishi, S. Seino, S., and Bell, G.I. (1988) Biochem.
Biophys. Res. Commun. 157, 937-943.

Schwab, D.A., and Wilson, J.E. (1989) Proc. Natl. Acad.
Sci. USA 86, 2563-2567.

Arora, K.K., Fanciulli, M., and Pedersen, P.L. (1990) J.
Biol. Chem. 265, 6481-6488.

Griffin, L.D., Gelb, B.D., Wheeler, D.A., Davison, D.,
Adams, V., and McCabe, E.R.B. (1991) Genomics 11, 1014-
1024.

Thelen, A.P., and Wilson, J.E. (1991) Arch. Biochem.
Biophys. 286, 645-651.

Schwab, D.A., and Wilson, J.E. (1991) Arch. Biochem.
Biophys. 285, 365-370.

Deeb, S.S., Malkki, M., Laakso, M. (1993) Biochem.
Biophys. Res. Commun. 197, 68-74.

Bork, P., Sander, C., and Valencia, A. (1993) Prot. Sci.
2, 31-40.

Kopetzki, E., Entian, K.-D., and Mecke, D. (1985) Gene
39, 95-102.

Stachelek, C., Stachelek, J., Swan, J., Botstein, D., and
Konigsberg, W. (1986) Nucleic Acids Res. 14, 945-963.

Colowick, S.P. (1973) in The Enzymes (Boyer, P.D., Ed.),
3rd ed., Vol. 9, pp. 1-48, Academic Press, New York.

Easterby, J.S., and O’Brien, M.J. (1973) Eur. J. Biochem.
38, 201-211.

Ureta, T. (1982) Comp. Biochem. Physiol. 718, 549-555.

White, T.K., and Wilson, J.E. (1989) Arch. Biochem.
Biophys. 274, 375-393.

Chothia, C., and Lesk, A.M. (1986) EMBO J. 5, 823-826.

64

17.

18.

19.

20.

21.

22.

230

240

25.

26.

270

28.

29.

30.

310

32.

65

Harrison, R. (1985) Crystallographic Refinement of Two
Isozymes of Yeast Hexokinase and Relationship of
Structure to Function. Ph.D. thesis, Yale University, New
Haven, CT.

Baijal, M., and. Wilson, J.E. (1992) Arch. Biochem.
Biophys. 298, 271-278.

Arora, K.K., Filburn, C.R., and Pedersen, P.L. (1991) J.
Biol. Chem. 266, 5359-5362.

Arora, K.K., Filburn, C.R., and Pedersen, P.L. (1993) J.
Biol. Chem. 268, 18259-18266.

Chen, L., Zeng, C., and Fromm, H.J. (1993) FASEB J. 7,
A1176.

Magnani, M., Bianchi, M., Casabianca, A., Stocchi, V.,
Daniele, A., Altruda, F., Ferrone, M., Silengo, L. (1992)
Biochem. J. 285, 193-199.

White, T.K., and Wilson, J.E. (1987) Arch. Biochem.
Biophys. 259, 402-411.

White, T.K., and Wilson, J.E. (1990) Arch. Biochem.
Biophys. 277, 26-34.

Hutny, J., and Wilson, J.E. (1990) Arch. Biochem.
Biophys. 283, 173-183.

Ellison, ‘W.R., Lueck, J.D., and Fromm, H.J. (1974)
Biochem. Biophys. Res. Commun. 57, 1214-1220.

Ellison, W.R., Lueck, J.D., and Fromm, H.J. (1975) J.
Biol. Chem. 250, 1864-1871.

Chou, A.C., and. Wilson, J.E. (1974) Arch. Biochem.
Biophys. 165, 628-633.

Purich, D.L., and Fromm, H.J. (1971) J. Biol. Chem. 246,
3456-3463.

Lueck, J.D., and Fromm, H.J. (1974) J. Biol. Chem. 249,
1341-1347.

Sambrook, J., Fritsch, E.F., and Maniatis, T. (1989)
Molecular Cloning: A laboratory Manual, 2nd ed., Cold
Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

Bird, P., Gething, M.-J., and Sambrook, J. (1987) J; Cell
Biol. 105, 2905-2914.

33.

34.

350

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

66
DeWitt, D.L., El-Harith, E.A., Kraemer, S.A., Andrews,
M.J., Yao, E.F., Armstrong, R.L., and Smith, W.L. (1990)
J. Biol. Chem. 265, 5192-5198.

Housley, P.R., and Forsthoefel, A.M. (1989) Biochem.
Biophys. Res. Commun. 164, 480-487.

Wilson, J.E. (1989) Prep. Biochem. 19, 13-21.

Perrella, F.W. (1988) Anal. Biochem. 174, 437-447.

Hill, H.D., and Straka, J.G. (1988) Anal. Biochem. 170,

203-208.

Smith, A.D., and Wilson, J.E. (1991) Arch. Biochem.
Biophys. 287, 359-366.

Dixon, M., and Webb, E.C. (1964) Enzymes, Academic Press,
New York.

Grossbard, L., and Schimke, R.T. (1966) J. Biol. Chem.
241, 3546-3560.

Katzen, H.M., and Schimke, R.'T. (1965) Biochem. Biophys.
Res. Commun. 54, 1218-1225.

Gonzalez, C., Ureta, T., Babul, J., Rabajille, E., and
Niemeyer, H. (1967) Biochemistry 6, 460-468.

Wilson, J.E., and Smith, A.D. (1985) J. Biol. Chem. 260,
12838-12843.

Rose, I.A., Warms, J.V.B., and Kosow, D.P. (1974) Arch.
Biochem. Biophys. 164, 729-735.

Wilson, J.E., and Chung, V. (1989) Arch. Biochem.
Biophys. 269, 517-525.

Wilson, J.E. (1991) in Methods of Biochemical Analysis
(Suelter, C.H., Ed.), Vol. 35, pp. 207-250, Wiley-
Interscience, New York.

Wilson, J.E. (1978) Arch. Biochem. Biophys. 185, 88-99.
Wilson, J.E. (1979) Arch. Biochem. Biophys. 196, 79-87.
Wilson, J.E. (1982) Arch. Biochem. Biophys. 218, 254-262.

Smith, A.D., and Wilson, J.E. (1991) Arch. Biochem.
Biophys. 291, 59—68.

67

51. Smith, A.D., and Wilson, J.E. (1992) Arch. Biochem.
Biophys. 292, 165-178.

FOOTNOTES
‘Abbreviations used: Glc-6-P, glucose 6-phosphate; AnGlc-6-P,

1,5-anhydroglucitol 6-phosphate; SDS, sodium.dodecyl sulfate;
ANOVA, analysis of variance.

68

Chapter III
Functional Organization of Mammalian Hexokinases:
Both N- and C-Terminal Halves of the Rat Type II Isozyme

Possess Catalytic Sites

Archives of Biochemistry and Biophysics

Vol.329, No. 1, May 1, pp. 17—23, 1996

69

ABSTRACT

Previous ‘work. has shown that. catalytic function is
associated exclusively with the C-terminal half of the Type I
isozyme of mammalian hexokinase. In contrast, we now
demonstrate that both halves of the Type II isozyme possess
comparable catalytic activities. Mutation of a catalytically
important Ser residue to Ala at analogous positions in either
the N- or C-terminal halves (8155A or S603A, respectively) of
the rat Type II isozyme resulted in approximately 60%
reduction in specific activity of the enzyme, with more than
90% reduction in the doubly mutated enzyme (8155A/S603A).
Catalytic activity was retained in a chimeric hexokinase
comprised of the N-terminal half of Type II hexokinase and
catalytically inactive (by site-directed.mutation) C-terminal
half of the Type I isozyme. The N- and C-terminal catalytic
sites of Type II hexokinase are similar in Vmu and Km (£130
pH) for Glc; however, the N-terminal site has a lower (0.45 mM
vs. 1.1 mM) Km for ATP, is slightly 'more sensitive to
inhibition by the product analog, 1, 5-anhydroglucitol-6-P, and
is much more sensitive to inhibition by P,. It is suggested
that the Type II isozyme most closely resembles the 100 kDa
hexokinase which resulted from duplication and fusion of a
gene encoding an ancestral 50 kDa hexokinase and.which.was the
precursor for the contemporary Type I, Type II, and Type III
mammalian isozymes. Subsequent evolutionary changes could then

have led to functional differentiation of the N- and C-

70

71
terminal halves, as seen with the Type I (and possibly the

Type III) isozyme.

72

Mammalian tissues contain three isozymes of hexokinase
(ATP:D-hexose 6-phosphotransferase, EC 2.7.1.1), generally
designated as Types I, II, and III. These exist as monomeric
species with molecular weights of approximately 100,000 and
show similarity in amino acid sequence (reviewed in Ref. 1)
consistent with their membership in a closely related family
which also includes other kinases (2). Moreover, these
isozymes exhibit internal sequence repetition, i.e. there is
extensive sequence similarity between the N-terminal and C-
terminal halves of the enzymes (1,3,4), and these in turn are
similar to sequences of 50 kDa hexokinases such as those found
in yeast (5,6) or Schistosoma mansoni (7). Such observations
have led to the current view that the 100 kDa mammalian
isozymes have evolved by duplication and fusion of a gene
encoding an ancestral 50 kDa hexokinase, which also gave rise
to the 50 kDa hexokinases of contemporary organisms (1).

Site-directed.mutagenesis studies (8-12) have identified
several amino acid residues as being catalytically important,
and all of these are conserved in both N- and C-terminal
halves of all Type I, II, or III isozymes whose sequences have
been determined (1). Despite this, it is clear that the two
halves of the Type I isozyme are functionally distinct, with
catalytic activity associated solely with the C-terminal half
(9-11,13) while the N-terminal half is catalytically inactive
and thought to serve a regulatory function (13-15). Whether

this same functional organization exists in the Type II and

73
Type III isozymes has been uncertain.

A recent study (15) of chimeric hexokinases, constructed
by interchange of cDNA.segments encoding the:N- and C-terminal
halves of the Type I and Type II isozymes followed by
expression in C051 cells, confirmed that catalytic function
was associated with.the C-terminal half of Type II hexokinase,
as with the Type I isozyme. A chimera comprised of the
catalytically inactive N-terminal half of Type I hexokinase
and the C-terminal half of the Type II isozyme was
catalytically active. However, we also noted that higher
specific activities were associated with hexokinases
possessing the N-terminal half of the Type II isozyme (either
the Type II isozyme itself or the chimera produced by
combination of the N-terminal half of Type II with the C-
terminal half of Type I). As one possible explanation for this
observation, we speculated that, unlike the N-terminal half of
the Type I isozyme, the N-terminal half of the Type II isozyme
might retain intrinsic catalytic activity. The present study
has shown this to be the case. In addition to contributing to
our ‘understanding' of ‘the functional organization. of the
mammalian isozymes of hexokinase, these findings have
implications with respect to the evolutionary relationship

between these isozymes.

MATERIALS AND METHODS

Materials. Restriction enzymes, T4 DNA ligase, and Glc-6-
]? dehydrogenase 'were purchased from Boehringer' Mannheim
(Indianapolis, IN) and other biochemicals from Sigma Chemical
Co. (St. Louis, MO). Plasmid DNAs were purified using plasmid
purification kits obtained from Qiagen, Inc. (Chatsworth, CA).
DMEM (high glucose) for culture of €081 cells was a product of
GIBCO BRL (Gaithersburg, MD), and supplemented with fetal
bovine serum (2%) and defined bovine calf serum (8%) from
HyClone Laboratories (Logan, UT). The BCA Protein Assay
Reagent and BSA standard, as well as the Imject Activation
Immunogen Conjugation Kit were purchased from Pierce Chemical
Co. (Rockford, IL).

Site-direCted mutagenesis and construction of vectors for
expression in 0031 ceiie. A previously described (15) full-
1ength cDNA encoding rat Type II hexokinase was used. Sequence
encoding the N-terminal half of the enzyme is encoded by an
Eco RI-Nco I fragment, while a complementary Nco I-Pst I
fragment encodes the C-terminal half. This division 'was
convenient for selectively generating site-directed mutations
within the N- or C-terminal halves, and then reconstructing
full-length coding sequences with any desired mutations.

Site-directed mutagenesis was done using the Altered
Sites kit and pSelect vector from Promega (Madison, WI), as

previously described in detail (10). An Nco I site had been

74

75
engineered (10) into the multiple cloning site of the pSelect
vector to facilitate manipulations. The Eco RI-Nco»I or.Nco I-
Pst I fragments were cloned into the modified pSelect to give
vectors containing sequence encoding the N- or C-terminal
halves, respectively. Mutations were made at a Ser residue
that is conserved in both‘ N- and C-terminal halves of the
mammalian hexokinases (1) and which was previously shown to be
of critical importance for catalytic function in the Type I
isozyme (8,10). Mutated forms of the rat Type II isozyme with
Ser converted to Ala in the N-terminal half, the C-terminal
half, or both halves are designated as $155A, S603A, or
SISSA/S603A, respectively. The mutant 8155A was made using the
oligonucleotide:GGTTTCACCTTCQCGTTCCCCTG, where the underlined
base corresponds to conversion of the Ser codon (ECG) to the
Ala codon (QCG) ; the analogous mutation in the C-terminal
half, S603A, was made using the oligonucleotide
GGTTTCACATTCQCCTTCCCTTG, with the underlined base resulting in
conversion of the TCC coding for Ser to QCC coding for Ala.
Complete coding sequences for the Type II isozyme, with any
desired mutations, were reconstructed by ligation of
complementary fragments and cloning in pUC18. The construct
was then subcloned into pSVT7 for expresssion in COSl cells as
previously described (15). All mutations were confirmed by
sequencing, first in the pSelect vector in which the mutation
had been made, and then again after transfer of the mutated

construCt to pSVT7.

76

Expression of Type II hexokinase and mutants in COSl
eelle. The transfection procedure was exactly as described
previously. Sham transfected cells were treated identically
except that no plasmid DNA was added. Cells were harvested 3
days after transfection and extracts prepared as previously
described (15) except that the sonicate was centrifuged at
15,000 x g for 10 min prior to analysis. Protein and
hexokinase activity were determined immediately after
preparation of the extracts, which were then stored at -80° C
for further use.

Determination of hexokinase activity and protein.
Hexokinase activity was determined using a spectrophotometric
assay in which Glc-6-P formation is coupled to NADPH
production, monitored at 340 nm, in the presence of excess
Glc-6-P dehydrogenase (16). The Kms for substrates, Glc and
ATP, were determined under the same conditions, except for
appropriate variations in the concentration of the relevant
substrate. Prior to determination of the Km for Glc, Glc
present in the extraction buffer was removed by chromatography
on spin columns of Sephadex G-25 (fine), equilibrated with 50
mM Hepes, 0.5 mM EDTA, 10 mM thioglycerol, pH 7.5; absence of
Glc in the eluted enzyme was confirmed by enzymatic assay. All
kinetic data conformed to Michaelis-Menten behavior and were
analyzed using the EZ Fit program of Perrella (17) . Inhibition
by the Glc-6-P analog, AnGlc-6-P, and by P,and its analogs,

sulfate and arsenate, was determined as in previous studies

77
(10,15).

Protein was assayed using the BCA Protein Assay reagent.
Samples were pretreated with iodoacetamide (18) to avoid
interference from ‘thioglycerol present. in ‘the extraction
buffer.

Eroduction of a pelyclonal antiserum specific for the
Tyg II isozyme of hexokinase. While the isozymes of mammalian

 

hexokinase exhibit extensive similarity throughout virtually
all of their amino acid sequences, their C-terminal sequences
are unique (1). An octapeptide having the sequence CIREAGQR
was synthesized in the Macromolecular Structure, Sequencing &
Synthesis Facility, Michigan State University, with purity and
identity confirmed by MALDI TOF mass spectrometry. The last
seven residues in this peptide correspond to the C-terminal
sequence of the rat Type II isozyme of hexokinase, which is
quite distinct from that of the Type I or Type III isozymes;
the Cys residue at the N-terminus was included to facilitate
conjugation to carrier protein for immunization. The latter
was.done, with.keyhole limpet hemocyanin.as carrier, using the
Imject Conjugation Kit from Pierce Chemical Co. The conjugate
was provided to University Laboratory Animal Resources,
Michigan State University, for production of antiserum. A
rabbit was immunized with 100 pg of the conjugate, emulsified
in Hunter's Titer Max, with subsequent booster injections
approximately one and two months later. Titer was determined

by ELISA, essentially as described previously (19), except

78
that wells of the microtiter plate were coated with BSA that
had been conjugated with the above peptide, again using the
conjugation kit from Pierce Chemical Co. Several bleedings
showing good titer were combined and used for these
experiments.

SDS gel electrophoresis and impunoblottipg. Procedures
for SDS gel electrophoresis were essentially as described
earlier (11,20) except that blotting was done using a Trans-
Blot SD Transfer Cell (BioRad Laboratories, Richmond, CA).
Immunoblots were analyzed quantitatively (11) using the GDS-
2000 gel documentation system and associated software from
UVP, Inc. (San Gabriel, CA).

geguence comparisons. Sequences of the N- and C-terminal
halves of the isozymes were compared using the BESTFIT routine
of the GCG Sequence .Analysis Software Package (Genetics
Computer Group, Inc., Madison, WI). Default weighting values
(gap penalty = 3.00; gap extension penalty = 0.10) were used.
Sequences used in comparisons were: residues 1-475, 1-475, and
1-488 for the N-terminal halves of Types I, II, and III,
respectively; residues 476-918, 476-917, and 489-924 for the

respective C-terminal halves.

RESULTS

Expreesien of Zyg II hexokinase and mutant forms in cos;

cells. Specific activities of hexokinase in extracts from C081

cells transfected with pSVT7 constructs encoding the wild type

79

Type II isozyme or mutant forms are shown in Table I. As in
the previous study (15), transfection with the cDNA for the
wild type isozyme resulted in hexokinase activities that were
well above those in sham tranfected cells, i.e., above
endogenous levels of hexokinase in COS1 cells. Transfection
with cDNAs encoding the singly mutated forms, SISSA or S603A,
gave activities that were also well above those in sham
transfected cells, but lower than that with wild type cDNA.
Hexokinase activities in extracts from cells transfected with
cDNA encoding the double mutant, 8155A/8603A, were markedly
reduced.but still higher than those of sham transfected cells.

Examination of extracts by SDS gel electrophoresis (Fig.
1) disclosed the expected presence of an additional component
in extracts from transfected cells, migrating slightly ahead
of the Type I hexokinase (16) used as marker; no other
differences were detected between transfected cells and the
sham transfected control. It has previously been noted (15)
that, despite the close similarity in actual molecular weight
of the Type I and Type II isozymes, the Type II isozyme
migrates slightly faster on SDS gels. It was not possible to
quantitate the relative amounts of these expresSed species by
densitometric methods due to inadequate resolution from other
closely adjacent components. However, visual estimation of the
relative staining intensities indicated that the mutant forms
(both single and double mutant) were present in amounts

somewhat.greater than the*wild type enzyme. This was supported

80

Table I. Hexokinase Activity in Extracts of C081 Cells
Transfected with cDNAs Encoding Wild Type and Mutant Forms of
Type II Hexokinase

cDNA used for Hexokinase activity‘
transfection (unitslmg protein)
Sham control ‘ 0.05 i 0.02
Wild type 0.76 I 0.25
SlSSA . 0.52 i 0.16
S603A 0.51 i 0.21
$155A/8603A 0.09 i 0.03

 

'IMean 1 SD for 11 different transfections. ANOVA analysis and
t-test indicated significant differences (p<0.01) between sham
control and all other values, between wild type and all
mutants, and between the single mutants and the double mutant;
there was no significant difference between the S155A and
8603A mutants.

81

 

Figure 1. SDS-gel electrophoretic analysis of C081 cell
extracts. Lane 1, Type I hexokinase purified from rat brain
(16); lane 2, extract from sham-transfected C081 cells; lanes
3, extract from C081 cells transfected with cDNA encoding wild
type Type II hexokinase; lanes 4-6, extracts from C081 cells
transfected with cDNAs encoding mutants 8155A, 8603A, and the
double mutant 8155A/S603A, respectively. The gel was stained
with Coomassie Blue. The position of the Type II isozyme, and
mutant forms, is indicated by the arrow at the right.

82

by immunoblotting results (Fig. 2) which indicated. more
intense staining of bands corresponding to the mutant forms.
Since the epitope recognized by the antiserum is the C-
terminal sequence common to the wild type and.mutant forms, it
is reasonable to assume that the immunoreactivity of the
various forms is equivalent, and thus that more intense
staining on the blot is indicative of higher levels of the
particular protein.

Relative specific activities of wild type and mutant
forms of Type II hexokinase. Staining intensity of the Type II
hexokinase band on immunoblots increased linearly with the
amount of extract from cells transfected with the wild type
enzyme (Fig. 3). Since it is reasonable to assume that mutant
and wild.type forms exhibit equivalent.immunoreactivity, it is
therefore possible to estimate the relative amounts of wild
type and mutant proteins based on relative staining
intensities. On this basis, the mutants 8155A, 8603A, and
8155A/S603A were expressed at levels 1.5- to 2-fold higher
than the wild type enzyme.

Knowing the relative amounts of these enzymes present in
the extracts (from immunoblotting), as well as the specific
activities of hexokinase present in the extracts (e.g., Table
I), it ‘was possible to calculate the relative specific
activities of the mutant and wild type forms. The specific
activity of the endogenous C081 enzyme in cell extracts was

assumed to be constant and equal to that of the sham

83

Figure 2. Immunoblot analysis of C081 cell extracts. Extracts
were from: lane 1, sham-transfected cells; lanes 2-5, cells
transfected with cDNAs encoding wild type Type II hexokinase,
8155A, 8603A, or 8155A/8603A, respectively. The blot was
probed with antiserum raised against a peptide corresponding
to sequence at the C-terminus of rat Type II hexokinase.

84

 

600 °

I

400 °

200

Area (pixels)

 

 

 

0 20 40 60 80 100

Cell Extract (ill)

Figure 3. Densitometric analysis of an immunoblot. Increasing
amounts of an extract from C081 cells transfected with cDNA
encoding the wild type Type II isozyme were used, and the
intensity of the bands on the immunoblot was quantitated as
described in Methods. The resulting standard curve was then
used to estimate quantities of the mutant forms, 8155A, 8603A,
or 8155A/8603A, present in extracts from cells transfected
with the respective cDNAs; the latter extracts were included
on the same blot used for generating the standard curve.

85
transfected control; this value was subtracted from. the
'specific activities seen in extracts from cells transfected
with wild type or mutant cDNAs. The specific activity of a
mutant form, relative to that of the wild type enzyme, could
then be calculated as the ratio of the activity present in the
"mutant" extract and that present in a "wild type" extract
containing an equivalent amount of the expressed isozyme. On
this basis, the specific activities of the 8155A, 8603A, and
8155A/S603A mutants were calculated to be 36%, 43%, and 4% of
the wild type isozyme, respectively. In another determination,
using extracts from a different transfection experiment, the
corresponding values were quite similar, being 33%, 40%, and

7%, respectively.

Kinetic parameters and refllatory properties of wild type

and mutant forms of Tyg II hexokinase. The Kms of the
expressed wild type enzyme as well as the 8155A and 8603A

mutants are shown in Table II; the various forms were not
distinguishable in. their apparent affinity for' Glc, but
differed significantly in apparent affinity for ATP. The
mutant and wild type isozymes did not differ greatly in their
sensitivity to inhibition by AnGlc-6-P (Fig. 4), although
8155A was slightly less susceptible, and S603A slightly more
susceptible, than the wild type Type II isozyme. In contrast,
the 8155A mutant was markedly less sensitive, and the 8603A
mutant considerably more sensitive than wild type Type II

hexokinase to inhibition by P,(Fig. 5), previously shown to

86

Table II. Kinetic Properties of Wild Type and Mutant Forms of

Type II Hexokinase

Extract from cells
transfected with cDNA for

Wild type 146
8155A 126
S603A 143

i
i
i

Glucose (uMl'

22
15
10

Km for

ATP

0.64
1.09
0045

mM "

i
i
i

0.04
0.20
0.05

 

‘Mean i SD for three determinations, each with an extract from

a different transfection.

bMean f SD for six determinations, each with an extract from

a different transfection.

87

 

100
80

29
.g 60
O
4 40
s

 

20

 

 

 

1,5-Anhydroglucitol (uM)

Figure 4. Inhibition of wild type and mutant forms of Type II
hexokinase by the Glc-6-P analog, 1,5-anhydroglucitol-6-P.
Activity was determined as a function of increasing inhibitor
concentrations, with initial ATP concentration in each assay
equal to the Km value (Table II) for the respective enzyme.

(. ) , Wild type Type II hexokinase; (CI) , 8155A; (0) , 8603A.

‘88

 

100
80

 

z.»
.5 60-
0
<1 40
s

20

 

 

 

Phosphate (mM)

Figure 5. Inhibition of wild type and mutant forms of Type II
hexokinase by inorganic phosphate. Activity was determined as
a function of increasing phosphate concentrations, with
initial ATP concentration in each assay equal to the Km value
(Table II) for the respective enzyme. (0) , Wild type Type II
hexokinase; (D), 8155A; (0), 8603A.

89
inhibit competitively vs. ATP (21). Similar results (not
shown) were seen for inhibition by the P,analogs, sulfate and
arsenate.

Retention of activity in a chimeric hexokinese.

constrUcted from an inactivated C-terminal (catalytic) domaip
of Type I.and the N-terminal half’of Type II.hexokinase. It is

apparent that the most straightforward interpretation of the
above results is that both N- and C-terminal halves of Type II
hexokinase possess intrinsic, and comparable, catalytic
activities. Thus, mutation of a catalytically important Ser
residue in either half results in reduction of specific
activity to somewhat less than half of wild type, while the
double mutation results in almost complete loss of activity.
This was further examined by an alternative strategy based on
previous work from this laboratory (10,15).

Baijal and Wilson (10) showed that mutation of Ser 603 to
Ala resulted in virtually complete loss of catalytic activity
in the Type I isozyme. Using the same strategy previously used
to prepare chimeric enzymes from complementary halves of the
Type I and Type II isozymes (15), a chimera was prepared in
which the N-terminal half of Type II was combined with a C-
terminal half of Type I, but with Ser 603 changed to Ala in
the latter (10). It.has previously been shown (10,15) that the
wild type Type I isozyme, the Type I mutant S603A, and the
NIICI chimera (comprised of the N-terminal half of Type II

hexokinase and the C-terminal half of the Type I isozyme) are

90
all expressed at similar levels under these conditions. This
was confirmed in the present study and, based on staining
intensity' 'with Coomassie Blue, the mutated chimera
NIICI(8603A) was also expressed in amounts comparable to the
other forms (results not shown). Activities seen in extracts
of C081 cells transfected with these constructs are shown in
Table III; similar results were seen in other experiments
employing various subsets of these cDNAs. In confirmation of
previous reports, these results demonstrate the devastating
effect of the 8603A.mutation on activity of the Type I isozyme
(8,10) and the enhanced activity seen with the NIICI chimera
(15), the latter being the observation that led to the present
study (see comments in Introduction). Most interesting in the
present context is the last entry in Table III, demonstrating
retention of substantial catalytic activity in the chimeric
enzyme produced from the N-terminal half of Type II and an
inactive C-terminal half of Type I hexokinase. This clearly
supports assignation of catalytic activity directly to the N-

terminal half of the Type II isozyme.
DISCUSSION

The catalytic sites in the N- and C-terminal halves of
the Type II isozyme appear to function more-or-less
independently' - for» example, there is no indication of

cooperativity in binding of substrates. Moreover, they have

91

Table III. Hexokinase Activity in Extracts from C081 Cells
Transfected with cDNAs Encoding Wild Type and Chimeric Forms
of Hexokinase

 

cDNA used for Hexokinase Activity
transfection lunits/mg protein)
Sham 0.05

Type I 1.14

Mutant Type I(S603A) 0.07

NIICI chimera 1.87

NIICI(8603A) chimera 0.66

 

92
comparable V5“ values, with mutation of analogous
catalytically important residues (Ser 155 or Ser 603) in
either half having similar effect on specific activityfl. The
two sites do differ significantly in Km for ATP but not for
Glc, with the catalytic site in the N-terminal half having the
higher apparent affinity for this nucleotide substrate. In
principle, differences in KIn should lead to deviation from
monophasic Michaelis-Menten kinetics, but this would not be
expected to be detectable with the relatively slight
difference in the Kms of the N- and C-terminal sites for ATP
(22). Indeed, simulation of kinetic results that would be
expected from two sites functioning with the same V,m but with
Khs for ATP of 1.09 and 0.45 mM (Table II) was consistent with
monophasic Michaelis-Menten behavior and an apparent K, of
0.66 mM, virtually identical with the value found for the Type
II isozyme (Table II).

Deviation from simple Michaelis-Menten behavior might
also be predicted for the single mutants, 8155A and 8603A,
since they were comprised of N- and C-terminal halves with
markedly different Vmax values (22). However, this Ser to Ala
mutation results in reduction of Vpu to approximately 5% of
Vm for the unmodified enzyme (8,10, and present work).
Simulation demonstrates that the kinetics would be dominated
by the unmodified site, consistent with the observation of

monophasic Michaelis-Menten behavior for the mutant enzymes.

93

Okazaki et al. (23) have previously reported that the C-
terminal half of Type II hexokinase was preferentially labeled
by the ATP analog, 2’,3'-dialdehyde ATP. These authors took
this to indicate that the catalytic site was associated solely
with the C-terminal half, as with the Type I isozyme (9-
11,13). The present’ study indicates that an alternative
explanation is necessary,‘ perhaps a difference in .the
effectiveness with which this analog labels the ATP binding
sites in the N- and C-terminal halves of Type II hexokinase.
This is not unreasonable since the differences in Km for ATP

do, in fact, indicate that the two sites are not identical.
The two sites also <differ in ‘their sensitivity' to
inhibition by P,, with the N-terminal site having greater
affinity for this effector. Type I hexokinase also has two
sites for binding of P” with the higher affinity site again
being associated with the N-terminal half (14, 15,24) . However,
functionally, the effects of binding to these N-terminal sites
are quite different. With the Type II isozyme, binding results
in inhibition. In contrast, with the Type I isozyme, binding
of P,to the high affinity site in the N-terminal half does not
lead to inhibition, but rather, antagonizes the binding of
Glc-6-P and.thereby the inhibition.by this product; in effect,
binding of P, at the N-terminal site serves an activating
function with.the Type I isozyme. It has previously been noted
(21,25) that these different responses to the physiologically

relevant modulator, P“ are likely to be a significant factor

94
in defining distinct metabolic roles for these two isozymes.

As depicted schematically in Fig. 6, the 100 kDa
mammalian hexokinases appear to have evolved by duplication
and fusion of a gene for an ancestral 50 kDa hexokinase
(1,13). This 50 kDa ancestral hexokinase had already acquired
sensitivity to inhibition by the product, Glc-6-P, and thus
resembled the hexokinase found in present-day organisms such
Pas starfish (13) or the parasitic S. mansoni (26). The initial
100 kDa. hexokinase produced in this fashion 'would. have
possessed two catalytic sites, one in each half and both
subject.to regulationlby'the product Glc-6-P, as found for the
Type II isozyme. We thus suggest that the Type II isozyme is
the more "ancient", most closely resembling the ancestral
hexokinase first generated by the gene duplication and fusion
event. Subsequent evolutionary changes resulted in
differentiation of function in the N- and C-terminal halves,
as found for the Type I isozyme. The functional organization
of the Type III isozyme remains to be determined.

The N- and C-terminal halves of Type II hexokinase
exhibit 60% identity in amino acid sequence. The comparable
values for the Type I and Type III isozymes are 51% and 47%.
The greater similarity between the halves of the Type II
isozyme is also consistent with this isozyme being most
closely related to the ancestral 100 kDa product of gene
duplication and fusion (for which there would have been,

presumably, 100% identity between the halves). The lower value

95

 

+ -|e—
Yeut /
1e— u—
l Starfish
8. meat
Type II
/ \
1' 1 7 ‘1‘ IV
796 7 YPG
\\s
Typo in

Figure 6. Evolution of the isozymes of mammalian hexokinase.
Relative molecular weights (approximately 50,000 or 100,000)
are represented. by the length of the horizontal lines.
Catalytic sites are represented by filled circles, and
regulatory sites, at which Glc-6-P is bound with resulting
inhibition, are represented by filled squares. Hexokinases are
viewed as evolving from an ancestral 50 kDa hexokinase,
insensitive to regulation by Glc-6-P, similar to the enzyme
found in present-day yeast. Acquisition of a regulatory site
resulted in an ancestral 50 kDa, Glc-6-P-sensitive form whose
descendants persist in present day organisms such as starfish
and S. mansoni. Duplication and fusion of the gene encoding
this ancestral form gave rise to the 100 kDa, Glc-6-P-
sensitive mammalian isozymes. The initial product of this
duplication and fusion event would be expected to possess two
catalytic sites, each sensitive to regulation by Glc-6-P. As
shown in the present study, Type II hexokinase has these
characteristics and accordingly, is suggested to most closely
resemble this ancestral form. The Type I isozyme evolved by
further mutations that resulted in functional differentiation
of the two halves, the C-terminal half retaining catalytic
function while the 'N-terminal half took on additional
regulatory roles; in the course of the latter process, the Glc
binding site in the N-terminal half, and the Glc-6-P binding
site in the C-terminal half, of the Type I isozyme became
latent (indicated by the open symbols) - see refs. 1, 13, or
24 for more extensive comments about the functional
organization. of the Type I isozyme. The disposition of
catalytic and regulatory functions within the Type III isozyme
remains to be determined. Although the mammalian Type IV
isozyme ("glucokinase") is a 50 kDa enzyme, like the
hexokinases of yeast, starfish, and S. mansoni, sequence
comparisons indicate that it is much more closely related to
the 100 kDa mammalian isozymes. Thus, Type IV hexokinase is_
suggested (28,31) to have evolved by a resplitting of the
gene, with additional mutations resulting in loss of
sensitivity to Glc-6-P.

96

for the Type I isozyme is consistent with further mutations
giving rise to the known functional differentiation in this
isozyme. Based on the still lower identity between the two
halves of Type III hexokinase, we predict that functional
differentiation will also be found between the halves of the
Type III isozyme (though not necessarily the same as that seen
with the Type I isozyme).

A fourth isozyme, Type IV hexokinase, more commonly
called. glucokinase, is also found in. mammalian tissues,
principally liver and pancreas but also in selected cells in
brain and gut (27). In contrast to Types I-III, the molecular
weight of Type IV hexokinase is about 50,000, and it is
insensitive to inhibition by product Glc-6-P (1). Based on
sequence comparisons (1,2,28) , Type IV hexokinase is much more
closely related to the 100 kDa mammalian hexokinases than it
is to the 50 kDa hexokinases found in yeast or S. mansoni;
moreover, there is conservation of the intron/exon structure
in the genes for Types II and IV hexokinase (29,30), further
implying very close relationship between these two species. We
thus agree with the suggestions of Ureta (31) and Griffin et
al. (28) that the Type IV hexokinase most likely arose from a
resplitting' of the gene encoding’ a 100 kDa "mammalian"
hexokinase (Fig. 6), rather than evolving independently from
an ancestral 50 kDa hexokinase that was the precursor for the
100 kDa enzymes (29,30).

After submission of this manuscript, we learned of the

97
recently published results of Ardehali et al. (32) , which are
in agreement with the present study, i.e. , catalytic activity
is associated with both N- and C-terminal halves of Type II
hexokinase. (We are grateful to Dr. Daryl Granner for
providing us with a preprint of this publication.) Ardehali et
al. (32) also found that, relative to the N-terminal site, the
C-terminal site had a somewhat higher KIll for ATP and decreased
sensitivity to inhibition by Glc-6-P. However, the magnitude
of these differences was considerably greater than observed in
the present work, probably owing to differences in kinetic
methods used. It is of particular note that Ardehali et al.
(32) expressed the individual halves of the Type II isozyme in
catalytically active form, demonstrating that - as previously
shown for the Type I isozyme (9,12,13) - catalytic function

does not require interactions with the complementary half.

110.

341.

342.

1J3.

141.

155.

165,

17'.

REFERENCES

Wilson, J.E. (1995) Revs. Physiol. Biochem. Pharmacol.
126, 65-198.

Bork, P., Sander, C., and Valencia, A. (1993) Prot. Sci.
2, 31-40.

Schwab, D.A., and Wilson, J.E. (1991) Arch. Biochem.
Biophys. 285, 365-370.

Thelen, A.P., and Wilson, J.E. (1991) Arch. Biochem.
Biophys. 286, 645-651.

Kopetzki, E., Entian, K.-D., and Mecke, D. (1985) Gene
39, 95-102.

Stachelek, C., Stachelek, J., Swan, J., Botstein, D., and
Konigsberg, W. (1986) Nucl. Acids Res. 14, 945-963.

Shoemaker, C.B., Reynolds, S.R., Wei, G., Tielens,
A.G.M., and Harn, D.A. (1995) Exp. Parasitol. 80, 36-45.

Arora, K.K., Filburn, C.R., and Pedersen, P.L. (1991) J.
Biol. Chem. 266, 5359-5362.

Arora, K.K., Filburn, C.R., and Pedersen, P.L. (1993) J.
Biol. Chem. 268, 18259-18266.

Baijal, M., and. Wilson, J.E. (1992) Arch. Biochem.
Biophys. 298, 271-278.

Baijal, M., and Wilson, J.E. (1995) Arch. Biochem.
Biophys. 321, 413—420.

Zeng, C., and Fromm, H.J. (1995) J. Biol. Chem. 270,
10509-10513.

White, T.K., and Wilson, J.E. (1989) Arch. Biochem.
Biophys. 274, 373-393.

Hutny, J., and Wilson, J.E. (1990) Arch. Biochem.
Biophys. 283, 173-183.

Tsai, H.J., and.‘Wilson, J.E. (1995) Arch. Biochem.
Biophys. 316, 206-214.

Wilson, J.E. (1989) Prep. Biochem. 19, 13-21.

Perrella, F.W. (1988) Anal. Biochem. 174, 437-447.

98

18.

19.

20.

21.

22.

230

24.

25.

26.

27.

28.

29.

30.

31.

32.

99

Hill, H.D., and Straka, J.G. (1988) Anal. Biochem. 170,
203-208.

Finney, K.G., Messer, J.L., DeWitt, D.L., Wilson, J.E.
(1984) J. Biol. Chem. 259, 8232-8237.

Polakis, P.G., and Wilson, J.E. (1984) Arch. Biochem.
Biophys. 234, 341-352.

Lueck, J.D., and Fromm, H.J. (1974) J. Biol. Chem. 249,
1341-1347.

Dixon, M., and Webb, E.C. (1964) Enzymes, Academic Press,
New York.

Okazaki, H., Takebayashi, Y., Ando, M., Date, 8., Tokuda,
H., and Ishibashi, S. (1992) Mol. Cell. Biochem. 117, 87-
92.

White, T.K., and Wilson, J.E. (1990) Arch. Biochem.
Biophys. 277, 26-34.

Wilson, J.E. (1984) in Regulation of Carbohydrate
Metabolism (Beitner, R., Ed.), Vol. I, pp. 45-85, CRC
Press, Boca Raton, FL.

Tielens, A.G.M., van den Heuvel, J.M., van Mazijk, H.J.,
Wilson, J.E., and Shoemaker, C.E. (1994) J. Biol. Chem.
269, 24736-24741.

Jetton, T.L. , Liang, Y. , Pettepher, C.C. , Zimmerman,
E.C., Cox, P.G., Horvath, K., Matschinsky, F.M., and
Magnuson, M.A. (1994) J. Biol. Chem. 269, 3641-3654.

Griffin, L.D., Gelb, B.D., Wheeler, D.A., Davison, D.,
Adams, V., and McCabe, E.R.B. (1991) Genomics 11, 1014-
1024.

Printz, R.L., Koch, 8., Potter, L.R., O’Doherty, R.M.,
Tiesinga, J.J., Moritz, 8., and Granner, D.K. (1993) J.
Biol. Chem. 268, 5209-5219.

Kogure, K., Shinohara, Y., and Terada, H. (1993) J3 Biol.
Chem. 268 8422-8424.

Ureta, T. (1982) Comp. Biochem. Physiol. 713, 549-555.
Ardehali, H., Yano, Y., Printz, R.L., Koch, 8.,

Whitesell, R.R., May, J.M., and Granner, D.K. (1996) J.
Biol. Chem. 271, 1849-1852.

FOOTNOTES

‘Abbreviations used: Glc-6-P, glucose 6-phosphate; AnGlc-6-P,
1,5-anhydroglucitol 6-phosphate; DMEM, Dulbecco's modified
Eagle’s medium; Hepes, N-(2-hydroxyethyl)-1-
piperazineethanesulfonic acid; MALDI TOF, matrix assisted
laser desorption ionization time of flight; ELISA, enzyme-
linked immunosorbant assay; BSA, bovine serum albumin; SDS,
sodium dodecyl sulfate; ANOVA, analysis of variance.

2If both sites had exactly the same Va“ and were completely
independent, one would obviously expect mutations in either
half to reduce the specific activity to approximately 50%.
This is close to, but not identical with, the 60-70% reduction
estimated experimentally. Whether this discrepancy reflects
errors in the estimation, or greater complexity in the effect
of mutation, or both, is uncertain.

100

Chapter IV
Functional Organization of Mammalian Hexokinases:
Characterization of the Rat Type III Isozyme and Its
Chimeric Forms, Constructed with the N- and C-Terminal

Halves of the Type I and Type II Isozymes
Accepted for Publication in

Archives of Biochemistry and Biophysics

November, 1996

101

ABSTRACT

Previous studies have shown that catalytic function is
associated. with both halves of the Type II isozyme of
mammalian hexokinase, while the Type I isozyme is functionally
differentiated into a catalytic C-terminal half and regulatory
N-terminal half. The Type III isozyme has now been shown to be
similar to the Type I isozyme in its functional organization.
Chimeras comprised of the N-terminal half of Type III
hexokinase and the C-terminal half of either Type.I or Type II
hexokinase have activities that can be attributed to the C-
terminal half, and are similar in activity to chimeras
comprised of the C-terminal half of Type III and the
intrinsically inactive N-terminal domain of Type I or the
inactivated (by site-directed mutation) N-terminal half of
Type II hexokinase. Virtually no activity was seen with
chimeras constructed with the N-terminal half of the Type III
isozyme and catalytically inactive (by site-directed mutation)
C-terminal halves of Type I or Type II hexokinase. Substrate
inhibition by Glc is seen only with the Type III isozyme and
with chimeric forms containing the C-terminal half of Type III
hexokinase and the N-terminal half of Type I or Type II
isozyme, the latter inactivated by site-directed mutation;
this is attributed. to conformational changes induced. by
binding of Glc to a low affinity site in the N-terminal half,
with.subsequent effect.on.catalyticlactivity'of the C-terminal

half. These results also provide further insight into the role

102

 

103
of interactions (or lack of interactions) between the N- and
C-terminal halves in the inhibition of the Type I-III isozymes
by Glc-6-P, its antagonism by low concentrations of P,, and the

inhibition seen at higher concentrations of Pr

T

104

Three isozymes of hexokinase (ATP:D-hexose 6-
phosphotransferase, EC 2.7.1.1) with molecular weights of
approximately 100,000 are found.in mammalian.tissues (see Ref.
1 for review). The Type I, Type II, and Type III isozymes
differ in their tissue distribution, in their tendency to
associate with specific subcellular structures (Types I and II
with mitochondria, Type III with the nuclear periphery), and
in their kinetic and regulatory properties. It is reasonable
to presume that these:differences adapt the three isozymes for
distinct roles in mammalian glucose metabolism.

Despite their differences, the 100 kDa mammalian isozymes
also exhibit notable similarities. For example, all three
isozymes are sensitive to inhibition by physiologically
relevant levels of the product, Glc-6-T”, and this is
generally considered to be a major factor in regulation of
activity in vivo. In addition, all three isozymes show
striking similarity throughout their amino acid sequence.
Moreover, there is internal sequence repetition, i.e., in.each
isozyme, the sequences of the N-terminal and.G-terminal halves
are similar, both to each other and to the sequences of 50 kDa
hexokinases found in lower eukaryotes such as yeast or
Schistosoma mansoni. Thus, the 100 kDa mammalian isozymes are
thought to have evolved by duplication and fusion of a gene
encoding an ancestral 50 kDa hexokinase.

The initial product of a gene duplication and fusion

event would be expected to be a 100 kDa enzyme consisting of

105
two halves that are functionally equivalent. It is now clear
that this is the case for the Type II isozyme (2,3). Both the
N- and C-terminal halves have been shown to possess catalytic
activity that is responsive to inhibition by Glc-6-P. In
contrast, the two halves of the Type I isozyme are
functionally distinct, with catalytic activity associated
solely with the C-terminal half (4-7) while regulatory
function, i.e., the site at which the inhibitory Glc-6-P is
bound, has been ascribed to the N-terminal half (4,8,9)2.‘The
Type II isozyme has been proposed (3) to be the more
"ancient", i.e., to most closely resemble the original 100 kDa
hexokinase generated by duplication and fusion of the gene
encoding the ancestral 50 kDa form. Duplication of the gene
encoding this initial 100 kDa form would then give rise to
additional 100 kDa isozymes ‘which could undergo 'further
mutations leading to functional differentiation of the N- and
C-terminal halves. As noted previously (3) , the sequence
similarity - 60% identity - between the N- and C-terminal
halves of the Type II isozyme is more extensive than the 51%
identity in sequences of the two halves of Type I hexokinase.
Moreover, when the three isozymes are compared, there is
considerably' greater' conservation. of sequence in ‘the C-
terminal halves than there is in the N-terminal halves (17).
These observations are consistent with the view that
functional differentiation of the Type I isozyme resulted from

mutations in the N-terminal half, while greater sequence

 

106

conservation, with preservation of catalytic function,
occurred.in the C-terminal half. Based on even less similarity
- 47% identity - between the sequences of the N- and C-
terminal halves of the Type III isozyme, it was predicted (3)
that this isozyme, like the Type I isozyme, would exhibit
functional differentiation between its N- and C-terminal
halves.

The functional organization of rat Type III hexokinase
has been determined in. the present study. In addition,
chimeric constructs, comprised of the N- or C-terminal half of
Type III hexokinase and the complementary half from the Type
I or Type II isozyme, have been expressed and characterized.
The results provide additional insight into the role of
interactions between the N- and C-terminal- halves in
determining the kinetic and regulatory properties of the

mammalian isozymes, Types I-III.

MATERIALS AND METHODS

Materials. Restriction enzymes, T4 DNA ligase, and yeast
Glc-6-P dehydrogenase were purchased from Boehringer Mannheim
(Indianapolis, IN). NADP was from United States Biochemical
(Cleveland, OH) , and other biochemicals were products of Sigma
Chemical (St. Louis, MO). Site-directed mutations were done
using the Altered Sites Kit and pSelect vector from Promega

(Madison, WI). For DNA sequencing, the Sequenase Version 2.0

107

DNA Sequencing Kit from United States Biochemical was used.
The pcDNA3 expression vector was from Invitrogen (San Diego,
CA). Plasmid DNAs were purified using‘ kits from. Qiagen
(Chatsworth, CA). Ham's F-12 medium and DMEM, as well as the
transfection reagent, lipofectAMINE, were products of Gibco
BRL (Gaithersburg, MD). Sera used for cell culture were
purchased from HyClone Laboratories (Logan, UT). The BCA
Protein Assay Reagent and BSA standard were from Pierce
Chemical (Rockford, IL). Antibodies used for immunoblotting -
polyclonal (rabbit) antiserum against the Type I (18) and‘Type
II (3) isozymes, monoclonal antibodies 5A.and 21 reacting‘with
the'Type I isozyme (19), and.monoclonal antibody C7C3 specific
for the Type III isozyme (20) - have previously been
described.

Determination of hexokinase activity and protein.
Hexokinase activity and protein were determined exactly as
previously described (9,21). Briefly, hexokinase activity was
determined by coupling Glc-6-P formation to NADPH production,
monitored at 340 nm, in the presence of excess Glc-6-P
dehydrogenase. Kinetic parameters were determined under
standard assay conditions except for variation in the
concentration of the relevant substrate; the Kms for glucose
and ATP determined in this way were actually apparent Kms but
the distinction is not critical in this context and, for
brevity, we refer to them simply as "Km". Kinetic data were

analyzed using the EZ-Fit program of Perrella (22) . Inhibition

 

108
by P,cr the Glc-6-P analog, 1,5-AnGlc-6-P, were determined as
in our previous studies (3,9); both of these inhibitors are
competitive vs ATP.

Protein 'was determined with the BCA Protein Assay
Reagent, with BSA as standard; samples were pretreated with
excess iodoacetamide (23) to avoid interference from the
thioglycerol present in the samples.

Electrophoretic methods and immunoblotting. Procedures
for SDS gel electrophoresis and immunoblotting, and for non-
denaturing electrophoresis on cellulose acetate plates were as
previously described (3,9).

Construction of plasmids for the expression of Type III
hexokinase and its chimeric forms. It was convenient to
generate separate plasmids with inserts encoding either the N-
terminal or C-terminal.halves.of'the isozymes. Construction of
pUC18 plasmids containing N- or C-terminal halves of the Type
I and Type II isozymes has been described (3,6,7,9); in
addition, previous work (3) has provided constructs of the
Type II isozyme in which catalytically important (6,24) Ser
residues in the N-terminal (Ser155) or C-terminal (Ser603)
half were mutated to Ala, resulting in inactivation of the
catalytic sites in the respective halves. All of these
constructs were designed so that sequences encoding N-terminal
halves were cloned as Eco RI-Nco I fragments, while sequences
encoding the C-terminal halves were cloned as Nco I-Pst I

fragments. Complete coding sequences for the 100 kDa isozymes

 

109
or chimeras could then be obtained by ligation of the
individual halves at the central Nco I site. A similar
strategy was employed in the present work, but was somewhat
more complicated due to the presence of additional Eco RI and
Nco I sites within the coding sequence for the Type III
isozyme (25).

The cDNA encoding the rat Type III isozyme had been
cloned in pUC18 as two separate Eco RI fragments (25),
comprised of nucleotides 1-525 and 526-3029 (numbering of
nucleotide positions refers to the composite sequence as
deposited in GenBank, Accession No. 073859). Plasmids
containing’ these inserts are: designated. p12.1 and. p4.1,
respectively.

The insert from p4.1 (nt 526-3029) was subcloned into the
Eco RI site of pSelect, generating the plasmid pSelect4.1.
Elimination of an Nco I site (nt 733) and the Eco RI site at
the 3’ end (nt 3029) and creation of an Nco I site at nt 1480
were performed simultaneously, following the protocol
suggested by Promega and using oligonucleotides having the
sequences ATGTGGTAGCIATGGTGAATG , GTTAGCAAGGQATTCGAGCTC , and
GGGGTGTGGCQATGGTGACTG, respectively; the underlined bases
indicate the mutations, which were designed to avoid any
‘change in amino acid sequence. Since these mutations resulted
in modification of restriction sites, clones were initially
selected based on restriction analysis, and mutations

subsequently confirmed. by DNA sequencing. The resulting

110

plasmid is designated pSelect4.1m. An Nco I- Xba I fragment,
comprised of nt 1481-3029 from. the Type III hexokinase
sequence plus about 20 additional bases at the 3' end that
corresponded to sequence between the modified Eco RI site and
the Xba I site in the multiple cloning region of pSelect,
represents the sequence encoding the C-terminal half of the
enzyme.

Digestion of pSelect4.1m with Eco RI and Nco I generated
a fragment corresponding to nt 526-1480, which was subcloned
into a previously modified (6) pSelect vector in which an Nco
I site had been introduced into the multiple cloning site.
After digestion of this vector with Eco ' RI, the Eco RI
fragment derived from p12.1, corresponding to nt 1-525, was
inserted to generate a construct containing the complete
coding sequence for the N-terminal half of Type III
hexokinase, designated pSelect-NIII. Restriction sites for Nco
I and Eco RI, at nt 250 and 525, respectively, were then
eliminated by site-directed mutations using oligonucleotides
TTCTGTGTTCIATGGAGCAGG and GCCTCTCTGAQTTCCTGGATG ; the
underlined bases indicate the mutations, which were without
effect on amino acid coding. Mutations were again verified by
restriction analysis and sequencing.

The complete cDNA.for Type III hexokinase, or for various
chimeric conStructs, was then generated in pUC18 by
combination of the appropriate N- or C-terminal fragments.

Identity of the complete constructs was confirmed by

 

111

restriction analysis, as in earlier studies (3,9). Following
the previous practice (3,9), chimeras are designated based on
the origin of their N- and C-terminal halves. For example,
NICIII is the chimera containing sequence encoding the N-
terminal half of Type I hexokinase and the C-terminal half of
the Type III isozyme. Other chimeras studied were NIIICI and
NIIICII. In addition, chimeras were constructed with the N- or
C-terminal halves of Type II hexokinase, inactivated by the
S155A or 8603A mutations (3), respectively, or with the C-
terminal half of Type I hexokinase inactivated by the S603A
mutation (6). The inactivated halves are designated as NII’,
CII', or CI’; the chimeras generated were NII'CIII, NIIICII',
and NIIICI’ .

Constructs were then subcloned into pSVT7 (26) or pcDNA3,
for expression in C081 (27) or M+R 42 (28) cells,
respectively.

Expression in C08 1 and M+R 42 cells. Initial experiments
in the present study were done using C081 cells, following
protocols used in our previous work (3,9). C081 cells express
the T antigen (27), and.when transfected with a vector such as
pSVT7, containing the SV40 early promoter, relatively high
levels of expression Igenerally' result. This has obvious
advantages and explains the popularity of C081 cells as an
expression system. However, a disadvantage in the use of C081
cells in the present context is the presence of endogenous

Ihexokinase activity (as would, of course, be true for most

 

112

eukaryotic cells). The similarity among mammalian hexokinases
makes it difficuLt to resolve the expressed enzyme from
endogenous hexokinase, especially when working on the scale
typical of transfection experiments. Although the endogenous
activity generally represents only a small fraction of the
total activity present (6), it nonetheless may complicate
interpretation of results, particularly those based on kinetic
methods.

The availability of M+R 42 cells (28), generously
provided by Dr. Michael Morgan, prompted us to examine
expression of hexokinase in this host. M+R 42 cells are a line
derived from Chinese hamster ovary (CHO) cells, and are
defective in expression of hexokinase (28). Initial
experiments indicated that, not unexpectedly, transfection of
M+R 42 cells with the constructs in pSVT7 did not lead to
appreciable expression of hexokinase activity. However,
transferring the constructs to the pcDNA3 vector, in which
expression is driven by the CMV promoter, resulted in levels
of expression similar to those seen in C081 cells.

M+R 42 cells were grown at 37° C and 5% C02 in Ham’s F-
12 medium supplemented with 4 mM sodium pyruvate, 2 mM
glutamine, and 10% fetal bovine serum or, in later
experiments, Cosmic Calf Serum (HyClone Catalog # A-2169); the
cells grew significantly faster with the latter serum.
Approximately 10° cells were plated per 100 mm tissue culture

dish, 18 hrs before transfection. The medium was removed by

 

113

aspiration and the cells rinsed with 4 ml PBS. The
transfecting DNA (5 pg) was mixed with 20 pl lipofectAMINE in
1 m1 of serum-free F-12 medium, and this was added to the
plate, followed by an additional 2 ml of F-12 medium (serum-
free). After incubation for 4-5 hrs at 37°, the transfection
medium was replaced by 10 ml of F-12 medium containing serum,
pyruvate, and glutamine (as above), followed by an additional
5 ml (of this medium 36-48 hrs later. Three days after
transfection, the medium was removed, cells rinsed with PBS,
and 11ml lysis buffer (50 mM Bicine, 10 mM thioglycerol, 10 mM
glucose, 0.5 mM EDTA, 0.1% (v/v) Triton X-100, pH 8.2) added.
After two freeze (-80°C)/thaw (room temperature) cycles, the
samples were microfuged for 10 min at 4° C and supernatants
removed for analysis.

Concentration and removal of glucose from cell extracts.
Prior to determination of the K, for Glc or studies of Glc
inhibition, glucose was removed from the samples by
centrifugation on spin columns of Sephadex G25 (Fine), as
jpreviously described (9). When concentration of extracts was
necessary (before removal of Glc on spin columns, and for
lmondenaturing electrophoresis), the extracts were centrifuged

Elt.45,000 rpm for 1 hr in a Beckman Ti50 rotor (4°C) prior to

concentration with an Centricon 30 device (Amicon, Beverly,

MA).

RESULTS

Expression in C081 and M+R 42 cells. Hexokinase
activities found in extracts of transfected C081 or M+R 42
cells are compared in Table I. It is evident that, using the
pcDNA3 vector, expression in M+R 42 cells resulted in levels
of activity comparable to those seen in C081 cells transfected
with the corresponding constructs in pSVT7. However, a
striking difference was the absence of significant hexokinase
activity in the sham transfected M+R 42 cells, consistent with
the reported defect in expression of endogenous hexokinase in
these cells (28). The results reported in the remainder of
this paper were obtained with isozymes or chimeras expressed
in M+R 42 cells.

The identities of various expressed constructs were
confirmed by immunoblotting (Fig. 1) . Thus, products expressed
from constructs containing the C-terminal half of Types I, II,
or III hexokinase are reactive with monoclonal antibody 5A,
the polyclonal antiserum against the C-terminal peptide of
Type II hexokinase, or monoclonal antibody C7C3, respectively.
Constructs including the N-terminal half of Type I hexokinase

led to expression products reactive with monoclonal antibody
2 1 . Moreover, since the same amount of protein was loaded in
each lane of the gel, the similarity in intensity of staining
on the immunoblots indicated that the various species were

eJ'tpressed at comparable levels. For example, the similar

114

115

Table I. Hexokinase Activity in Extracts of Transfected C081

and M+R 42 cells

cDNA used for Hexokinase Activity (units/mg protein)

 

 

 

transfection C081 celle‘ M+R 4; cellsb
Sham control 0.04 i 0.01 <0.001

Type I 0.23 i 0.06 0.16 i 0.05
Type II 0.66 i 0.21 0.37 i 0.14
Type III NDc 0.12 i 0.05
NICIII ND 0.11 i 0.04
NII’CIII ND 0.10 i 0.03
NIIICI 0.16 i 0.03 0.16 i 0.03
NIIICII 0.21 i 0.05 0.12 i 0.04
NIIICI’ ND 0.008 t 0.002
NIIICII’ ND 0.007 t 0.003
‘Constructs were in pSVT7 vector. Mean '1 SD for 4
transfections.

I’Constructs were in pcDNA3 vector. Mean 1 SD for 3
determinations with sham transfected cells or cells
transfected with NIIICI' or NIIICII' cDNAs; mean i SD for 15

‘transfections with other cDNAs.
cNot determined.

H
H

6

NII’CIII
N IIICI
NIIICII

NICIII

 

I
I

—.--——-.

2.“:
00
:1 C:
HH
ZZ

 

 

 

 

 

 

 

Figure 1. Characterization of expressed hexokinase isozymes
and chimeric forms by immunoreactivity. Left panel: replicate
immunoblots prepared after SDS gel electrophoresis of extracts
(90 pg protein) from sham transfected M+R 42 cells or from
cells transfected with the cDNAs indicated at the top of the
figure. The blots were probed with, top to bottom: monoclonal
antibody 5A, which recognizes an epitope in the C-terminal
half of Type I hexokinase (19); monoclonal antibody 21,
recognizing an epitope in the N-terminal half of Type I
hexokinase (19); polyclonal antiserum recognizing the C-
terminal sequence of Type II hexokinase (3); monoclonal
antibody C7C3 (20), recognizing an epitope in the C-terminal
half of Type III hexokinase. Monoclonal antibody 5A also
reacted with an endogenous component (intense bands in all
lanes in lower part of top panel) found in M+R 42 cells. This
is included in the figure as a convenient lane marker; the
uniform staining of this component also confirms consistency
in loading and blotting of the gel. Right panel: replicate
immunoblots prepared after SDS gel electrophoresis of extracts
(50 pg protein) from cells transfected with the cDNAs
indicated at the top of the figure. The blots were probed with
polyclonal antiserum against the Type I (top) or Type II
(bottom) isozymes.

117

intensities seen on the blot probed with monoclonal antibody
5A indicated that the Type I isozyme and the NIIICI chimera
were present in similar amount, while the relative staining
with monoclonal antibody 21 confirmed that this was also the
case with the Type I isozyme and the NICIII chimera. By the
same logic, the NICIII chimera was present in amounts
comparable to the Type III isozyme itself as well as the
NII’CIII chimera, as shown by the similar intensity of
staining on the blot.probed.with.monoclonal antibody C7C3. The
results shown at the right in Fig. 1 confirm that the NIIICI'
and NIIICII’ chimeras were expressed at levels comparable to
the 'Type I and. Type II isozymes themselves; thus, the
extremely low hexokinase activity in extracts from cells
transfected with the constructs encoding NIIICI’ and NIIICII’
(Table I) reflects the inactivity of the chimeric forms, not
a failure of expression.

Results of SDS gel electrophoresis (Fig. 2) were also
consistent with expression of the various constructs at
similar levels, although this was complicated by the presence
of substantial amounts of endogenous components which had
mobilities in the range of interest.3 Nonetheless, in some
cases, the expressed products could be clearly distinguished
(e.g., the Type I isozyme in Lane 3 and the NIIICI chimera in
Lane 8 of Fig. 2). In other cases, the expressed products
overlapped with an endogenous component of the M+R 42 cells,

but their presence could be inferred from increases in

 

118

u—4
r—r r—t
r— H r—t

Figure 2. SDS gel electrophoretic analysis of M+R 42 cell
extracts. First lane, Type I hexokinase, isolated from rat
brain (21) . The other lanes were extracts (90 pg protein) from
sham transfected cells or cells transfected with the cDNA
indicated at the top of the respective lanes. The gel was
stained with Coomassie blue.

NICIII

Q
E

'i NIIICII
1e

 

H
H
H
U
I
F—i
’2
w
I

       

 

119

staining intensity when compared with the corresponding band
seen with sham transfected cells. While this is admittedly
semi-quantitative, taken together with the blotting results
(Fig. 1), these results support the conclusion that the
various constructs are expressed at comparable levels in M+R
42 cells, and thus that the activity seen in cell extracts
(Table I) reflects the intrinsic activity of the expressed
isozyme or chimera.

Nondenaturing electrophoresis. All of the (active)
expressed species migrated as single components during
electrophoresis under nondenaturing conditions (Fig. 3). As
previously noted (9), the mobility of the various chimeric
forms is largely determined by the origin of the N-terminal
half.

Kinetic properties of the expressed isozymes and
chimeras. The Km'values for ATP and Glc are compared in Table
II. The Km values for Glc are identical to those previously
reported for the Type I and Type II isozymes expressed in C081
cells, while the Km values for ATP were slightly less than
those found in the previous study ( 3) . Whether the latter
represents experimental variation or reflects a contribution
from the endogenous hexokinase present in C081 cell extracts
is not clear. In any case, the results with the expressed Type
I, Type II, and Type III isozymes are consistent with earlier
work (29) indicating that the apparent affinity for Glc

increases in the order of Type II<Type I<Type III. The K, of

 

120

 

Figure 3. Electrophoresis of expressed isozymes or chimeric
forms under nondenaturing conditions. The first lane (at left)
contained.Type:I hexokinase, isolated from.rat brain (21). The
other lanes received extracts (approx. 5 milliunits of
hexokinase activity) from cells transfected with the cDNAs
shown at the top of the figure. After electrophoresis, the
cellulose acetate plate was stained for hexokinase activity
(9). The NIIICII chimera was relatively unstable under the
electrophoresis conditions and is barely perceptible in this
photograph. However, on the original plate, this was visible
as a distinct, though faint, band of activity with a mobility
just slightly greater than the NIIICI chimera.

121

Table II. KIn Values for Glucose and ATP‘

 

 

CDNA expressed Km for Glucose (11M) anor ATP (mM)
Type I 61 I 10 0.38 i 0.03
Type II 150 i 18 0.42 i- 0.01
Type III 33 I 4b 0.73 i 0.08
NICIII 36 i- 4” 0.36 i- 0.03
NII’CIII 37 i 8" 0.49 i 0.06
NIIICI 52 i 5 0.78 i- 0.08
NIIICII 155 i- 15 0.93 i 0.16
Wean '1 8D for three determinations with samples from

different transfections.

I"I‘hese hexokinases exhibit substrate inhibition (see text).
All data used for calculation of Km were obtained at Glc
Concentrations below those causing inhibition (< 0.5 mM) .

122
the expressed Type III isozyme for ATP was considerably higher
than that for the expressed Type I or Type II isozymes;
Grossbard and Schimke (29) also found the Type III isozyme to
have the highest Km for ATP.
The Type III isozyme exhibits substrate inhibition as the
Glc concentration is increased into the millimolar range
(29,30). This was also the case with the expressed Type III
isozyme and with the expressed chimeric forms, NICIII and
NII’CIII (Fig. 4). Substrate inhibition was not seen with the
isolated Type I or Type II isozymes (29) nor, in the present
study, with the expressed isozymes or NIIICI or NIIICII
chimeras (data not shown). Thus, substrate inhibition by Glc
is uniquely associated with the C-terminal half of Type III
hexokinase.
We also wish to note anomalous kinetic behavior observed
with the Type III isozyme and the chimeric forms showing
glucose inhibition, i.e., NICIII and NII’CIII. Whenmassayed.at
noninhibitory Glc concentrations (<0.5 mM), the reactions
:reached steady state within seconds of mixing the reaction
components. However, at inhibitory Glc concentrations, a
Inarked lag was observed before attainment of the steady state
rate. The length of the lag was related to the level of
inhibition, being on the order of 1-2 min at 2 mM Glc but 3-4
Ruin at 10 mM Glc. No lag was seen at any Glc concentration
With the Type I or Type II isozyme, nor with chimeric forms

not showing Glc inhibition. This anomalous behavior had not

 

123

 

 

 

100
80 I
a

60

% Vmax

40

20

 

 

Glucose (mM)

Figure 4. Substrate inhibition of the Type III isozyme and the
NICIII and NII’CIII chimeras by glucose. Activity of the Type
III isozyme (0), NICIII (:1), or NII’CIII (fl) is expressed
as a percentage of V,m values, determined (22) using only data
collected at Glc concentrations less than 0.5 mM. Each point
is the mean for two experiments. Error bars shown for the Type
III isozyme indicate the range of variation; where no error
bars are seen, the variations were smaller than the symbols.
For clarity, error bars are not shown for NICIII and NII'CIII,
but the variation was comparable to that for the Type III
isozyme. The curves are purely to guide the eye, and were
drawn using the "spline fit" option in SlideWrite Plus
(Advanced Graphics Software, Inc.); for clarity, no line has
been fitted to the data for NICIII.

124

been reported in previous studies of the Type III isozyme from
rat liver (29) or from rat Novikoff hepatoma cells (30).
However, an inquiry to Dr. Tito Ureta elicited the information
(T. Ureta, personal communication) that, though not mentioned
in the published report, this lag had been observed at
inhibitory Glc concentrations in the studies with the Type III
isozyme from Novikoff hepatoma cells (30) as well as with the
Type III from frog liver (31). Moreover, the lag was not
observed when fructose was used as a substrate for Type III
hexokinase from either Novikoff hepatoma cells (30) or frog
liver (31); neither of these enzymes shows substrate
inhibition by fructose. Thus, the appearance of the lag phase
is indeed correlated with the phenomenon of substrate
inhibition; moreover, this anomalous kinetic behavior is a
general property of the Type III isozyme, whether isolated
(30,31) or expressed.

Inhibition by the Glc-6-P analog, 1,5-AnGlc-6-P. When
assayed at 3.3 mM Glc, the concentration in the standard
assay, the Type I, Type II, and Type III isozymes. exhibited
similar sensitivity to inhibition by 1,5-AnGlc-6-P (Fig. 5) .
The chimeric forms, NICIII and NII’CIII, were similar to the
parental isozymes in this respect. In contrast, the NIIICI
Chimera was significantly more sensitive, and the NIIICII
chimera markedly less sensitive than the parental isozymes to
inhibition by 1,5-AnGlc-6-P. It is interesting to note that

1:lle chimeras constructed with the N-terminal half of Type III

125

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

E

100 I %
80 r lihi l l I
IE 60 * l iii-{7
3 40- E l1}
62
20 -
52% iii:

 

 

 

 

 

 

 

 

 

 

 

 

 

lmrcm

31 333431 331"” Eli
10 " 20 so 100
1,5 -Anhydroglucitol-6-P (pM)

 

Figure 5. Inhibition by 1,5-AnGlc-6-P at 3.3 mM Glc. Results
are expressed as percent of the activity seen in the absence _
01:5 inhibitor, and are the mean 1 SD for 4 determinations, each
w1th extracts from a different transfection.

126

hexokinase are thus similar to the analogous chimeras
constructed with the N-terminal halves of Type I or Type II
hexokinase (9), i.e., the NIICI chimera was much more
sensitive to inhibition, and the NICII chimera much less
sensitive than the parental isozymes to inhibition by 1,5-
AnGlc-6-P. Thus increased sensitivity to this inhibitor is
seen with chimeras of the form XCI, and decreased sensitivity
to inhibition is found with chimeras of the form XCII, where
X is the N-terminal half derived from either of the other
isozymes.

Since the Type III isozyme as well as the NICIII and
NII'CIII chimeras experience significant inhibition at 3.3 mM
Glc (Fig. 4), the sensitivity of these forms to inhibition by
1,5-AnG1c-6-P was also determined at. a: noninhibitory
concentration of 0.5 mM Glc. As shown in Fig. 6, this made a
substantial difference. Now the Type III isozyme was less
sensitive than the Type I and Type II isozymes to inhibition
by the Glc-6-P analog, consistent with previous reports
(29,30) that, when assayed at noninhibitory Glc
concentrations, the Type III was the least sensitive to
inhibition by Glc-6-P. Although the differences were less
striking with the chimeric forms, NICIII and NII'CIII, these
too were less sensitive to 1,5-AnGlc-6-P when assayed at the
reduced Glc concentration (Fig. 6). It is thus clear that
ignhibition of the Type III isozyme, or chimeras including the

CZ-terminal half of this isozyme, by Glc-6-P or its analog is

 

127

 

 

 

 

 

E

 

NHKIH

 

 

 

Type!

E

 

a
E

 

E

 

unxmn

 

 

 

T1901

:18
EE

 

E

 

 

NHKIH

 

 

 

11201

l

a
E

 

E

 

 

NHKIH

 

 

 

100
_} "I‘
80* pl—
3‘
"5' OUT
'5
o
< 40-
5°
20" ....BE
E65
0

10

20

50

100
1,5-Anhydroglucitol-6-P (pM)

Figure 6. Inhibitionlof Type III hexokinase and.the.NICIII and
NII’CIII chimeras by 1,5-AnGlc-6-P at 0.5 mM Glc. ReSults are
expressed as percent of the activity seen in the absence of
inhibitor, and are the mean 1 SD for 4 determinations,
with extracts from a different transfection. For comparison,

the results with the Type I and Type II isozymes at 3.3 mMIGlc'

(Fig. 5) have been included.

each

 

128

markedly influenced by the concentration of Glc present, with
elevated Glc enhancing inhibition by the 6-phosphate compound.
Although Radojkovic and Ureta (30) did not explicitly comment
on this, it is evident from the data in their Fig. 6 that they
observed a similar effect with the Type III isozyme from rat
Novikoff tumor cells, and.Ureta (31) had previously noted that
Glc-6-P was a potent inhibitor of the Type III isozyme from
frog liver "but only at high glucose levels".

Inhibition of the Type I isozyme, but not Type II or Type
III hexokinase, by Glc-6-P or its analogs is antagonized by
low concentrations of P, (1) . In the present study, the
reversal of inhibition of the expressed Type I isozyme was
again confirmed, but this effect was not seen with any of the
other isozymes or chimeric forms described here (data not
shown). For the NICIII chimera, this was examined at both 0.5
mM (noninhibitory) and at 3.3 mM (inhibitory) Glc, but no
reversal was seen in either case.

Inhibition by Pr 1% is also an inhibitor of the Type I,
Type II, and.Type III isozymes (1,9,12,32). This was confirmed
with the expressed isozymes in the present work (Fig. 7).
Although the Type I isozyme was considerably less sensitive,
all three isozymes showed substantial inhibition by low mM
levels of Pp The chimeric forms were similar to the Type II
and Type III isozymes in their sensitivity to this inhibitor,
although the NIIICII chimera was marginally less sensitive

than the other chimeras.

Ti

 

129

 

100r{_{_

80

PH
1+1
i-——}——I
I-l-I
Ill
I-—l-—I
I-l-I
Hj
H4
H...
lib!
I—q-—i
I-j—I

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.5 ”l
2 40-
8°
2: ".8384 mafia! 333% ii
1 , 2 5 10 .

A Phosphate (mM)

Figure 7. Inhibition by P, at 3.3 mM Glc. Results are expressed
as percent of the activity seen in the absence of inhibitor,

and are the mean 1 SD for 4 determinations, each with extracts .

from a different transfection.

130
The results shown in Fig. 7 were obtained with 3.3 mM Glc
in the reaction mixture. Reducing the Glc concentration to a
noninhibitory level (0.5 mM) did not affect the inhibition of
the Type III isozyme or the NICIII and NII'CIII chimeras by P,

(Fig. 8).
DISCUSSION

The specific activity of hexokinase in extracts of M+R 42
cells expressing the Type III isozyme (Table I) was comparable
to that obtained after expression of the Type I isozyme, and
was approximately half of the specific activity seen in
extracts of cells expressing the Type II isozyme, in which
both N- and C-terminal halves are known to possess catalytic
sites of essentially equal activity (2,3). Moreover,
expression of chimeras in which the C-terminal half of Type
III hexokinase was fused with N-terminal halves known to be
inactive (either the intrinsically inactive NI, or NII', which
had been inactivated by site-directed.mutation) gave specific
activities comparable to those seen with the Type I and Type
III isozymes. These reSults are consistent with the existence
of a single catalytic site, present in the C-terminal halves
of the Type I (4-7) and Type III isozyme. The intrinsic
specific activities of these catalytic sites are similar, to
each other and to those located in both N- and C-terminal

halves of the Type II isozyme (2,3); this is, of course, not

131

 

 

PH

ems

100 " FF}

 

 

 

 

 

 

 

 

 

 

IE - {r
8 60 lli
; 40- 9}
20-Ea BE 5 E J 5
ii a: a a E a

NICIII
NICIII
mcm
NII'CIII
Type III
Type III
NICIII
un'cm

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Tnnln
him!!!
NICIII

332:?

 

 

1 2 5 10
Phosphate (mM)

Figure 8. Lack of an effect of Glc concentration on the
inhibition of Type III hexokinase and the NICIII and NII'CIII
chimeras by P,. Results are expressed as percent of the
activity seen in the absence of inhibitor, and are the mean 1
SD for 4 determinations, each with extracts from a different
transfection. The results for each form are indicated by a
pair of bars; in each pair, the bar at the left indicates
results at 3.3 mM Glc (shown in Fig. 7, but included here for ‘
comparison) and the bar at the right indicates results with
0.5 mM Glc.

132

totally unexpected, considering the likely common origin of
these isozymes. Furthermore, chimeras in which the N-terminal
half of Type III hexokinase was combined with the
catalytically active C-terminal halves of Type I or Type II
hexokinase were expressed.with specific activities comparable
to those seen with the Type I and Type III isozymes
themselves, indicating the lack of significant contribution
from. NIII to the catalytic activity of these chimeras.
Confirming this, chimeras comprised of the N-terminal half of
the Type III isozyme and inactivated C-terminal halves from
the Type I or Type II isozymes, CI' and CII’, respectively,
were virtually inactive. We conclude that, as with the Type I
isozyme (4-7), catalytic function is associated solely with
the C-terminal half of the Type III isozyme. This conclusion.
is consistent with, and extends, the recent study of Palma et
al. (33); these authors reported that the C-terminal half of
human Type III hexokinase, expressed as a discrete species
from the truncated cDNA, was catalytically active.

The Kms of the chimeric forms for Glc are correlated with
the origin of the C-terminal half (Table II). This implies
that, at least with respect to binding of Glc at the catalytic
site, the N-terminal half of the molecule has essentially no
effect. In contrast, the Type III isozyme itself, as well as
chimeric forms in which the catalytically inactive NIII is
combined with catalytic C-terminal halves of Type I or Type II

hexokinase, have Kms for ATP that are similar and

 

133

significantly higher than those for the Type I or Type II
isozymes or for the .NICIII and. NII’CIII chimeras. This
correlation between the presence of an N-terminal half from
Type III hexokinase and a relatively high Km for ATP implies
the existence of substantial interactions between the N- and
C-terminal halves, with the N-terminal half from Type III
evoking an increased Km for ATP in the catalytic C-terminal
half, regardless of its source. other results, discussed
below, provide further evidence for a significant role of the
N-terminal half in regulation of catalytic function intrinsic
to the C-terminal half of Type III hexokinase, as previously
deduced for the Type I isozyme (4,8,10,15).

The expressed C-terminal half of Type III hexokinase, in
contrast to the Type III isozyme itself, did .not show
substrate inhibition by Glc, leading Palma et al. (33) to
propose that substrate inhibition might result from binding of
Glc to the N-terminal half of the intact 100 kDa isozyme. We
concur with this suggestion. Both N- and C-terminal halves of
the Type I and Type II isozymes contain binding sites for Glc
(2-4,15), although, in the intact 100 kDa Type I isozyme, the.
N-terminal site is thought to be latent (15). Based on the
extensive similarity between these isozymes, the presence of
a Glc binding site in the N-terminus of Type III hexokinase
can reasonably be inferred. Binding of Glc or G1c-6-P to sites
in the C- or N-terminal halves, respectively, has been shown

to induce marked conformational changes involving both halves

134

of mammalian Type I isozyme (1,4,8,15). Thus, it is clear that
extensive interactions between the halves can occur, with
resulting effect on structure and function. While Glc binds
with high affinity to the catalytic site in the C-terminal
half’of Type III hexokinase, conformational changes induced.by
binding of Glc to a lower affinity site in.the N-terminal half
could well exert a negative influence on catalytic function of
the C-terminal half.

Based on the above logic, and since the Type III isozyme
is unique in exhibiting inhibition by substrate Glc, it had
been anticipated that Glc inhibition would be correlated with
the presence of NIII in chimeric forms. In fact, that was not
the case. The NIIICI and NIIICII did not show substrate
inhibition, while the iNICIII and. NII’CIII chimeras did.
Clearly, substrate inhibition was correlated with the C-
terminal half of Type III hexokinase. Our interpretation of
these results is that the C-terminal half of Type III is
uniquely receptive to an inhibitory conformational "signal"
induced by binding of Glc to a site in the N-terminal half.
The N-terminal halves of any of the three isozymes are capable
of generating that signal. All residues known to be involved
in binding of Glc are conserved in the N-terminal‘ halves of
all three isozymes (1). It is not unreasonable to expect that
Glc is bound in a similar manner to, and with similar effect
on conformation of, the N-terminal halves of all three

isozymes. Glc inhibition is not seen with the Type II isozyme

135

because the Glc binding sites in both N- and C-terminal halves
are catalytically functional (2,3), i.e., binding to the N-
terminal half is productive, not inhibitory. Glc inhibition is
not seen with the Type I isozyme because the Glc binding site
in the N-terminal half is latent (15). A corollary of this
proposal is that, in contrast to the situation in the intact
Type I isozyme, the Glc binding site in the N-terminal half of
the NICIII chimera.is.not latent, presumably due to structural
differences between the native Type I and the NICIII
chimera.

The above analysis leads to the view that the functional
organization.of Type III hexokinase is the same as that of the
. Type I isozyme .(4,6,8,10,15), i.e., catalytic function is
associated with the C-terminal half and regulatory function
with the N-terminal half. We believe this extends to
inhibition by Glc-6-P and its analogs. Thus, we suggest that,
at low' Glc concentrations, these 6-phosphates bind. with
relatively low affinity to the N-terminal half of Type III
hexokinase, inducing a conformational change that results in
loss of the ability to bind ATP at the catalytic site in the
C-terminus (as inferred from the competitive nature of the
inhibition, implying mutually exclusive binding). In contrast
to the situation with the Type I isozyme (15), Glc can bind to
a site in the N-terminal half of the Type III isozyme. It has
previously been shown (4,12,15) that Glc and Glc-6-P (and

their analogs) bind synergistically to both the (isolated) N-

 

 

136

and C-terminal halves of Type I hexokinase, and the results
presented here demonstrate similar synergism with the Type III
isozyme, i.e., inhibition by 1,5-AnGlc-6-P is enhanced in the
presence of elevated (inhibitory) Glc levels. Thus, in the
case of the Type III isozyme, binding of Glc to the lower
affinity site in the N-terminal half 0f the molecule would
enhance binding of Glc-6-P to the regulatory site in the N-
terminal half, resulting in increased sensitivity to this
inhibitor. This is precisely what is seen with the Type III
isozyme (30,31, and present work). Furthermore, this
enhancement of inhibition is also seen with the NICIII and
NII'CIII chimeras which, as noted above, also exhibit
inhibition by the substrate, Glc. It is thus apparent that
both substrate inhibition as well as Glc-induced enhancement
of inhibition by Glc-6-P and its analogs can be explained by
the unique ability of the C-terminal half of Type III
hexokinase to respond to conformational signals induced by
binding of regulatory ligands to the N-terminal half of the
molecule.

It is not within the scope of the present report to
provide an.explanation for the anomalous kinetic results - the
marked lag prior to attainment of steady state in the presence
of inhibitory concentrations of Glc - seen with the Type III
isozyme and the NICIII and NII'CIII chimeras. It does seem
likely that this might involve a slow conformational response

to the binding of substrate Glc when the "inhibitor" site has

 

137
previously been occupied by a second Glc. At noninhibitory Glc
concentrations, no lag was seen with either of the other
inhibitors, P,cr 1,5-AnGlc-6-P.

If inhibition of Type I hexokinase resulted from Glc-6-P
or its analogs binding directly to the catalytic C-terminal
half, as some have proposed (5,14,16), it might be expected
that the response of chimeric forms to this inhibitor would be
more-or-less independent of the origin of the N-terminal half;
after all, other kinetic parameters, the Kms for Glc and ATP,
are only modestly affected, if at all, by the origin of the N-
terminal half (3,9, and present study). However, the present
work as well as our previous study (9) clearly demonstrates
that this is not the case. The markedly increased sensitivity
of chimeras of the form XCI and decreased sensitivity of XCII
chimeras, X being the N-terminal half derived from either of
the other isozymes, certainly suggests that inhibition by Glc-
6-P is not simply determined by the intrinsic sensitivity of
the catalytic C-terminal half. These results are, however,
consistent with the view that inhibition involves interactions
between the halves, with CI being more receptive and C11 less
receptive to conformational signals induced by binding of Glc-
6-P to a heterologous N-terminal half.

As noted above, inhibition of Type I hexokinase by Glc-6-
P or its analogs is antagonized by low concentrations of P“
This has previously been attributed to competition of Glc-6-P

(or its analogs) and P, for a common binding site in the N-

 

138
terminal half of Type I hexokinase (15); consistent with this,
15 also antagonized inhibition of the NICII chimera by 1,5-
AnGlc-6-P (9). Although similar antagonism between P,and 1,5-
AnGlc-6-P might have been expected with the NICIII chimera,
this was not Observed. However, this is readily explained by
the marked sensitivity of the Type III isozyme to inhibition
by P,. With Type III hexokinase, as with the Type I isozyme
(4,8,9,15), inhibition may be attributed to binding of P,tx:
the catalytic C-terminal half. Thus, sensitivity to P, is
independent of the nature of the N-terminal half, with the
Type III isozyme and the NICIII and NII'CIII chimeras being
equivalent in their response to this inhibitor, and
sensitivity to P, is also unaffected by binding of Glc (at
higher concentrations) to the inhibitory site in the N-
terminal half of these forms. Therefore, even though P,may in
fact be antagonizing the binding of Glc-6-P to the N-terminal
half of Type III hexokinase or the NICIII chimera, any
resulting increase in activity would be more than offset by
the direct, and rather potent, inhibitory effect of P, on
catalytic activity'of the C-terminal half. In contrast, the C-
terminal halves of the Type I (15) and Type II (3) isozymes
have relatively low affinities for P,. Thus, with Type I
hexokinase as well as the NICII chimera, significant
antagonism resulting from binding of P, to a high affinity site
in the N-terminal half, in competition with the inhibitory

hexose 6-phosphate, can be seen before being masked by

 

139
inhibition resulting from further binding of 1% to the low
affinity site in the C-terminal half.

The Type III isozyme is, like the Type I isozyme (4,6,8-
10,15), comprised of functionally distinct halves, catalytic
C-terminal and regulatory N-terminal. Thus, as previously
suggested (3), these isozymes likely represent the result of
duplication and further evolution of a gene encoding an
ancestral 100 kDa.hexokinase, to which the present-day Type II
isozyme remains most similar. The more complex regulatory
features seen with the Type I (e.g., P, antagonism of
inhibition by Glc-6-P) and Type III (e.g., ”substrate
inhibition. by' Glc) isozymes arise from close functional
interactions between the two halves. In contrast, the N- and
C-terminal halves of Type II hexokinase seem.to function more-
or-less independently (2,3). It is interesting that the major
evolutionary events leading to Type I and Type III isozymes
apparently rested.not so much.on modification.of affinities or
specificities of existing ligand binding sites, or development
of new ones, but rather, on development of functional
interactions between these sites with resulting acquisition of

‘unique catalytic and regulatory properties.

 

10.

11.

12.

13.

14.

15.

16.

REFERENCES

Wilson, J.E. (1995) Rev. Physiol. Biochem. Pharmacol.
126, 65-198.

Ardehali, H., Yano, Y., Printz, R.L., Koch, 8.,
Whitesell, R.R., May, J.M., and Granner, D.K. (1996) J.
Biol. Chem. 271, 1849-1852.

Tsai, H.J., and. Wilson, J.E. (1996) Arch. .Biochem.
Biophys. 329, 17-23.

White, T.K., and Wilson, J.E. (1989) Arch. Biochem.
Biophys. 274, 375-393.

Arora, K.K., Filburn, C.R., and Pedersen, P.L. (1993) J.
Biol. Chem. 268, 18259-18266.

Baijal, M., and. Wilson, J.E. (1992) Arch. .Biochem.
Biophys. 298, 271-278.

Baijal, M., and.‘Wilson, J.E. (1995) Arch. Biochem.
Biophys. 321, 413-420.

Hutny, J., and Wilson, J.E. (1990) Arch. Biochem.
Biophys. 283, 173-183.

Tsai, H.J., and. Wilson, J.E. (1995) Arch. .Biochem.
Biophys. 313, 206-214.

White, T.K., and Wilson, J.E. (1987) Arch. Biochem.
Biophys. 259, 402-411.

Ellison, ‘W.R., Lueck, J.D., and Fromm, H.J. (1974)
Biochem. Biophys. Res. Commun. 57, 1214-1220.

Ellison, W.R., Lueck, J.D., and Fromm, H.J. (1975) J.
Biol. Chem. 250, 1864-1871.

Chou, A.C., and ‘Wilson, J.E. (1974) Arch. .Biochem.
Biophys. 165, 628-633.

Magnani, M., Bianchi, M., Casabianca, A., Stocchi, V.,
Daniele, A., Altruda, F., Ferrone, M., and Silengo, L.

‘(1992) Biochem. J. 285, 193-199.

White, T K., and Wilson, J.E. (1990) Arch. Biochem.
Biophys. 277, 26-34.

Aleshin, A.E., Zeng, C., Fromm, H.J., and Honzatko, R.B.
(1996) FEBS Lett. 391, 9-10.

140

 

17.

180

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

141

Thelen, A.P., and Wilson, J.E. (1991) Arch. Biochem.
Biophys. 286, 645-651. '

Armstrong, R.L., Wilson,.J.E., and.Shoemaker, C.B. (1996)
Protein Expression and Purification, in press.

Wilson, J.E., and Smith, A.D. (1985) J. Biol. Chem. 260,
12838-12843.

Preller, A., and Wilson, J.E. (1992) Arch. Biochem.
Biophys. 294, 482-492.

Wilson, J.E. (1989) Prep. Biochem. 19, 13-21.
Perrella, F.W. (1988) Anal. Biochem. 174, 437-447.

Hill, H.D., and Straka, J.G. (1988) Anal. Biochem. 170,
203-208.

Arora, K.K., Filburn, C.R., and Pedersen, P.L. (1991) J.
Biol. Chem. 266, 5359-5362.

Schwab, D.A., and Wilson, J.E. (1991) Arch. Biochem.
Biophys. 285, 365-370.

Bird, P., Gething, M.-J., and Sambrook, J. (1987) J. Cell
Biol. 105, 2905-2914.

Gluzman, Y. (1981) Cell 23, 175-182.

Faik, P., Morgan, M., Naftalin, H.J., and Rist, R.J.
(1989) Biochem. J. 260, 153-155.

Grossbard, L., and Schimke, R.T. (1966) J. Biol. Chem.
241, 3546-3560.

Radojkovic, J., and Ureta, T. (1987) Biochem. J. 242,
895-903.

Ureta, T. (1976) J. Biol. Chem. 251, 5035-5042.

Lueck, J.D., and Fromm, H.J. (1974) J. Biol. Chem. 249,
1341-1347.

'Palma, F., Agostini, D., Mason, P., Dacha, M., Piccoli,

G., Biagiarelli, B., Fiorani, M., and Stocchi, v. (1996)
Mol. Cell. Biochem. 155, 23-29.

Arora, K.K., Filburn, C.R., and Pedersen, P.L. (1991) J.
Biol. Chem. 266, 5359-5362.

FOOTNOTES
‘Abbreviations used: Glc-6-P, glucose 6-phosphate; DMEM,
Dulbecco's modified Eagle medium; BSA, bovine serum albumin;
1, 5-AnGlc-6-P, 1, 5-anhydroglucitol 6-phosphate; PBS, phosphate
buffered saline (10 mM sodium phosphate, 154 mM NaCl, pH 7.0);
CMV, human cytomegalovirus.

2Theproposal that regulatory function is associated with the
N-terminal half of Type I hexokinase was initially based on
the demonstration that the N-terminal domain contained a
functional Glc-6-P binding site (10). Ligand binding studies
(11-13) had indicated the presence of a single Glc-6-P binding
site, and a single Glc binding site, in the intact 100 kDa.
enzyme. Thus it was reasonable to associate the binding of
Glc-6-P, and hence regulatory function, with the site in the
N-terminal half (10). This proposal has subsequently been
supported by additional evidence (8,9) . However, the situation
was complicated with the demonstration that the C-terminal
half, obtained either by limited proteolysis (4) or expressed
from the cDNA (5,14), retained sensitivity to inhibition by
Glc-6-P, demonstrating the presence of an intrinsic regulatory
site in the C-terminal half. A further complication was the
demonstration (4,15) that, when isolated, both N- and C-
terminal halves possessed binding sites for Glc as well as for
Glc-6-P. To reconcile these findings with the stoichiometry of
one Glc site and one Glc-6-P site per molecule of the intact
100 kDa enzyme (11-13), White and Wilson (15) proposed that,
in the intact enzyme, the Glc site in the N-terminal half as
well as the Glc-6-P site in the C-terminal half were "latent",
presumably'masked.by structural constraintSjpresent.in the 100
kDa enzyme; when these constraints were removed by separation
of the halves, both sites were revealed. Alternative
explanations certainly could be proposed but to be acceptable,
these must account for the existence of a functional Glc-6-P
binding site in the N-terminal half of Type I hexokinase (10),
and for the discrepancy between the existence of binding sites
for Glc-6-P and Glc in both N- and C-terminal halves, when
isolated, and the observed 1:1 stoichiometry for binding of
these ligands to the intact 100 kDa enzyme. Simple assertions
that it is the C-terminal site that is responsible for
inhibition of the intact enzyme by Glc-6-P (5,14,16) are
insufficient.

3In this respect, C081 cells possess an advantage since these
cells have relatively low levels of components running in this
region of the gel, and thus bands corresponding to the
expressed hexokinase isozymes, or chimeras, are readily
detectable (3,6,7,9).

142

 

4Although mutation of Ser155 to Ala, as in NII’ , or the
analogous Ser603 to Ala mutation in catalytic C-terminal half,
as in CI' or CII’, results in drastic reduction of catalytic

activity '(3,5,6,34), it does not abolish and may actually
enhance (34) binding of Glc.

143

Chapter V

CONCLUSION

144

 

145

Functional organization of mammalian 100 kDa hexokinases.
The objective of this study was to define the functional
organization in mammalian 100 kDa hexokinases. In certain
cases, several unique properties were found to be associated
with either the N- or C-terminal half ofthese enzymes. For
example, phosphate antagonism of inhibition by AnG-6-P
correlated with the N-terminal half of hexokinase I (1) when
the C-terminal.half’was substituted.with‘that.of hexokinase II
in the NICII chimera, and higher sensitivity to phosphate
inhibition correlated with the N-terminal half of hexokinase
II (1) when the C-terminal half was substituted with that of
hexokinase I in the NIICI chimera. However, there are also
results _that did not correlate with either half, e.g.
sensitivity to inhibition by the Glc-6-P analogue, AnG-6-P
(1,2). These properties are, presumably, due to the
interaction between the two halves.

The domain organization for catalytic function is not
uniform among all three mammalian hexokinases; hexokinase II
has one active site in each half of the protein (2), while
hexokinase I and III have only one active site in the C-
terminal halves (3,4). Because of the dual active sites and
greater conservation of protein sequence between the N- and C-
terminal halves (2) , hexokinase II functionally most resembles
the ancient hexokinase that has arisen from gene duplication
and fusion, which should have one active site in each of the

‘two identical halves (2). On the other hand, hexokinase I and

F1

 

146
III have only one functional active site, in their C-terminal
half, along with less conservation of amino acid sequence
between the N- and C-terminal halves; they have apparently
evolved further from hexokinase II (2,4).

Since the N-terminal halves of hexokinase I and III do
not possess catalytic function, a favored explanation is that
these N-terminal halves may have evolved into regulatory
domains (1-4). This speculation is consistent with the fact
that truncated.50 kDa C-terminal half of hexokinase I shows no
phosphate antagonism of Glc-6-P inhibition, while this
property is observed in the 100 kDa intact hexokinase I (3).
Although hexokinase I lost catalytic function in the N-
terminal half, the trade-off was the gain of a unique
regulatory function, providing hexokinase I with additional
regulatory sophistication (3).

A pew model for mammeliep 100 kDa hexokinases. The
hypothesis that duplication and fusion of a gene encoding an
ancestral 50 kDa hexokinase led to modern mammalian 100 kDa
hexokinases has become a well established and accepted model
(2). Because the duplication and fusion event occurred after
the formation of a discrete Glc-6-P site (3), physically,
there should be one Glc-6-P site and one glucose site in each
half of the 100 kDa mammalian hexokinase, i.e. two Glc-6-P
sites and two glucose sites on the entire molecule. However,
two separate laboratories (5,6) have reported that only one

glucose and one Glc-6-P site were detected on each 100 kDa

r

 

147
hexokinase I molecule. To explain this discrepancy, White and
Wilson (3) proposed a latent site model in which the glucose
site in the N-terminal half and the Glc-6-P site in the C-
terminal half were latent, presumably because of structural
impediment resulting from fusion of the two halves.

This latent site model, however, does not conform to the
dual active sites in hexokinase II (2), since the glucose site
in the N-terminal half of hexokinase II is obviously not
latent (it is catalytically active; therefore it must be
accessible to glucose). In addition, computer modelling of
mammalian hexokinases, based on the crystal structure of 50
kDa yeast hexokinase, does not exhibit obvious structural
impediment of the glucose binding site in the N-terminal half
(7). Thus, a revision of this model is needed.

One other model favored by Fromm (7), Pedersen (8), and
Magnani (9) is that the functional Glc-6-P site is located in
the C-terminal half of hexokinase I, where Glc-6-P competes
directly against substrate ATP (mutually exclusive). However,
this model cannot explain why Glc-6-P is able to protect the
N-terminal half of hexokinase I from trypsin proteolysis (3)
or sulfhydryl modification (10), and therefore this model
requires revision to accommodate such observations.

Developed with the concept that hexokinase II is the
closest descendant of the initial gene duplication and fusion
product, the new model is based on the assumption that N- and

C-terminal halves of hexokinases are similar structurally and

 

 

148
functionally. Known functional differentiations in the N-
terminal halves of hexokinase I and III can be explained with
derivations of this model. The following are descriptions of
the model:

1. The major assumptions of this model are that all three
mammalian hexokinases are similar in three-dimensional
structure, and that closure of the clefts (glucose binding
sites) in both N- and C-terminal halves occur in a concerted
manner because of the. structural connection. The former
assumption is based on the fact that all three 100 kDa
mammalian hexokinases are very similar in protein.sequence and
they have a common ancestor (2, Chapter 1 of this
dissertation). The latter assumption is essentially identical
to the symmetry model (11) first proposed.by Monod, Wyman, and
Changeux (12) where subunits of a protein can exist in two
different conformations, but all subunits in a protein
molecule adopt the same conformation. If one of the subunits
changes conformation, all subunits change conformation
simultaneously.» In other words, this is an extreme case of
co-operative interaction where the coefficient of co-
operativity is equal to the number of subunits. As a result
of this concerted conformational change in N- and C-terminal
halves, one molecule of glucose, when it binds to either half,
is sufficient to cause the closure of both N- and C-terminal
halves of hexokinase II, and thus prevents the binding of

another glucOse at the other empty active site. Because

 

 

149
probability of two glucose molecules binding to the two active
sites on the same hexokinase molecule simultaneously is very
small, the enzyme on average binds one glucose.at a given time
(Fig. 1).

This symmetry’ model would. predict that despite ‘the
presence of two active sites (glucose binding sites) in
hexokinase II, there will be only one detectable (accessible)
glucose binding site on every hexokinase II molecule. This
prediction is supported by the observation made by Connolly
and Trayer (13) that reaction with one molecule of glucose
analogue, N-bromoacetyl-glucosamine, is sufficient to cause
complete loss of catalytic activity in hexokinase II, despite
hexokinase II having two separate active sites. The
hypothetical mechanism is also supported by the observation
made in hexokinase I that despite the presence of two glucose
sites, demonstrable in the isolated halves, only one site is
detectable by binding assays (5,6).

Under the symmetry model, the ligand will bind to the
highest affinity site if the subunits have different
affinities for this ligand (see next two paragraphs). In the
case of glucose binding sites in hexokinase II, the N- and C-
terminal halves appear to have similar affinities for glucose
(2,14).

2. A similar mechanism is also true for Glc-6-P as an

inhibitory ligand. Glc-6-P binds to one of the two Glc-6-P'

sites closing both.Glc-6-P sites and exerts inhibitory effect

E1

 

1...:
'1

 

150

Figure 1. Proposed Symmetry Model for Catalysis by Mammalian
Type II Hexokinase.

   

 

 

GBP+ADP . ' GBPtADP

G: Glucose; A:ATP

a: Glucose analogues, N-acetyl glucosamine and N-bromoacetyl
glucosamine, cannot bind to hexokinase I at this step.

b: The N-terminal halves of hexokinase I and III cannot
catalyze the reaction at this step.

 

151

on both active sites. This hypothesis is based on a more
limited evidence from hexokinase I, that only one Glc-6-P site
is detectable on the entire 100 kDa molecule (5,6). Whether
hexokinase II and III are similar in this respect is yet to be
verified. Nonetheless, if this hypothesis is true, a single
Glc-6-P molecule should be sufficient to affect both active
sites in hexokinase II, which is supported by the fact that
the Glc-6-P analogue, AnG-6-P, is a competitive inhibitor to
substrate ATP (1), suggesting that both active sites can be
affected by this inhibitor.

It is noteworthy that the binding affinity for Glc-6-P
may not be identical in the N- and C-terminal halves. It has
been suggested that the N-terminal half of hexokinase II has
higher affinity for Glc-6-P than the C-terminal half (14). In
the presence of Glc-6-P, one molecule of Glc-6-P supposedly
binds to the high affinity site in the N-terminal half, which
is then sufficient to cause conformational change and lead to
inhibition at all active sites. Since binding of one Glc-6-P
molecule closes both.high and low affinity'Glc-6-P sites, only
the high affinity site will be detectable by binding assays
(5,6), which is why the K,of Glc-6-P for intact hexokinase II
is similar to the K, for the higher affinity site in the N-
terminal half (14).

Because of limited structural information about the Glc-

6-P binding site on hexokinases, there is no way of

 

152
distinguishing if both.Glc-6-P sites are equally accessible to
the ligand (symmetry model) or the site in the C-terminal half
is actually latent as proposed by White and Wilson (3).
Therefore, the latent model remains a viable alternative for
explaining the observation that only one Glc-6-P site is
detectable by the binding assays (5,6)

3. Because of amino acid substitutions in the N-terminal
half of hexokinase I, the active site in the N-terminal half
of hexokinase I has become inactive. However, glucose is
still able to bind to this site and causes closure of the
clefts (4). A similar scenario may also be true for the N-
terminal half of hexokinase III, since the N-terminal half of
hexokinase III possesses no catalytic activity but appears to
retain its ability to bind glucose, which results in substrate
glucose inhibition (4). Because of inactivity in the N-
terminal half of hexokinase I, 50 % of the cleft closure,
induced by the binding of substrates at the N-terminal half,
results in no catalysis. Meanwhile, every cleft closure in
hexokinase II results in catalysis regardless of which site
the catalysis actually occurs; therefore, the specific
activity of hexokinase II is about twice that of hexokinase I,
e.g. the specific activity of purified hexokinase I is about
60 U/mg protein (3,6,15,16), while the specific activity of
purified hexokinase II is about 100 U/mg protein (17).

4. Residue substitutions in the N-terminal half of

hexokinase I result in glucose analogues, N-acetyl glucosamine

 

 

 

153
and N-(bromoacetyl) -D-glucosamine being unable to fit into the
glucose binding site in the N-terminal half of hexokinase I.
The former analogue is unable to protect the N-terminal half
of hexokinaSe I from proteolytic attack (3), even when the N-
terminal half is in the 50 kDa truncated form (3), while the
latter analogue is unable to label (cross-link to) the N-
terminal half of intact hexokinase I (18). Nonetheless, both
analogues are able to bind at the active site in the C-
terminal half of hexokinase I, which explains why N-acetyl
glucosamine competes with glucose at the active site and
selectively protects the C-terminal half from proteolysis (3) ,
while N-(bromoacetyl)-D-glucosamine is able to affinity label

the C-terminal half of hexokinase I (18).

 

 

FUTURE RESEARCH

The presence of two glucose sites and two Glc-6-P sites
on each hexokinase I (3) and thepindication of only one
detectable glucose site and one Glc-6-P site per hexokinase I
molecule by binding assays (5,6) have not been adequately
explained on the structural basis. Whether this stoichiometry
remains true in hexokinase II and III should be the focus of
future research. Although Connolly and Trayer (13) had
demonstrated that one bromoacetyl-glucosamine per hexokinase
II molecule is sufficient to completely inhibit the hexokinase
activity, suggesting one detectable glucose site per 100 kDa
molecule, it is yet to be determined if the 1:1 stoichiometry
remains true with glucose as the ligand. If hexokinase II is
determined to have only one detectable glucose site despite
the fact of two functional active (glucose) sites, it would
imply that hexokinase II operates under the symmetry model as
proposed in this chapter.

It is obvious that the N-terminal halves of hexokinase I
and III do not possess catalytic function. However, according
to the gene duplication and fusion scheme, these halves should
have once been catalytically active, e.g. the N-terminal half
of modern hexokinase II still remains active. Like other
catalytically active halves in mammalian hexokinases, all
critical residues that are known to be involved in glucose
binding or catalysis are present in these two inactive halves

(19). Therefore, a logical question is what caused the loss

154

 

 

155

of catalytic function in these two halves? The answer to this
question will not only explain what happened in the evolution,
but also provide some insight into what other amino acids are
important for catalytic function in mammalian hexokinases.

Arora et al. (20) reported that mutation 8603A in
hexokinase I, as one may have expected, drastically hindered
the catalytic activity. Surprisingly, however, this mutation
increased the binding affinity for glucose. Although Harrison
(21) suggested that the 3-hydroxy1 group of glucose forms
hydrogen bond with the oxygen on the carbonyl group, not the
hydroxyl group on the side chain, of corresponding Ser 158 in
yeast hexokinase, it remains unanswered how the mutation can
affect the catalytic activity and at the same time increase
the affinity for glucose. A careful reexamination of this

increased affinity for glucose is needed.

10.

11.

12.

13.

14.

15.

16.

REFERENCES

Tsai, H.J., and Wilson, J.E. (1995) Arch. Biochem.
Biophys. 254, 316, 206-214.

Tsai, H.J., and Wilson, J.E. (1996) Arch. Biochem.
Biophys. 329, 17-23.

White, T.K., and Wilson, J.E. (1989) Arch. Biochem.
Biophys. 274, 375-393.

Tsai, H.J., and Wilson, J.E. (1996) Arch. Biochem.
Biophys. (in press).

Chou, A.C., and Wilson, J.E. (1974) Arch. Biochem.
Biophys. 163, 191-199.

Ellison, W.R., Lueck, J.D., and Fromm, H.J. (1975) J.
Biol. Chem. 250, 1864-1871.

Arora, K.K., Filburn, C.R., and Pedersen, P.L. (1993) J.
Biol. Chem. 268, 18259-18266. ‘

Solheim, L.P., and Fromm, J.H. (1981) Arch. Biochem.
Biophys. 211, 92-99.

Magnani, M., Bianchi, M., Casabianca, A., Stocchi, V.,
Daniele, A., Altruda, F., Ferrone, M., and Silengo, L.
(1992) Biochem. J. 285, 193-199.

Hutny, J., and Wilson, J.E. (1990) Arch. Biochem. Biophys.
283, 173-183.

Cornish-Bowden, A.J. (1976) Principles of Enzyme Kinetics,
Butterworths, London, England.

Monod, J., Wyman, J., and Changeux, J.-P. (1965) J. Mol.
Biol. 12, 88-118.

Connolly, E.A., and Trayer, I.P. (1979) Eur. J. Biochem.
93, 375-385.

Ardehali, H., Yano, Y., Printz, R.L., Koch, S., Whitesell,

R.R., May, J.M., and Granner, D.K. (1996) J. Biol. Chem.
271, 1849-1852.

Ellison, W.R., Lueck, J.D., and Fromm, H.J. (1974)
Biochem. Biophys. Res. Comm. 57, 1214-1220.

Jarori, G.K., Kasturi, S.R., and Kenkare, U.W. (1981)
Arch. Biochem. Biophys. 211, 258-268.

156

17.

18.

19.

20.

21.

157

Ganson, N.J., and Fromm, H.J. (1985) J. Biol. Chem. 260,
12099-12105.

Schirch, D.M., and Wilson, J.E. (1987) Arch. Biochem.
Biophys. 254, 385-396.

Wilson, J.E. (1995) Revs. Physiol. Biochem. Pharmacol.
126, 65-198.

Arora, K.K., Filburn, C.R., and Pedersen, P.L. (9991) J.
Biol. Chem. 266, 5359-5362.

Harrison, R.W. (1985) Ph.D. thesis, Yale University, New
Haven, CT.

"Ir1mmmmES