Báo cáo khoa học: Functional role of fumarate site Glu59 involved in allosteric regulation and subunit–subunit interaction of human mitochondrial NAD(P)+-dependent malic enzyme pptx

Functional role of fumarate site Glu59 involved in
allosteric regulation and subunit–subunit interaction of
human mitochondrial NAD(P)
+
-dependent malic enzyme
Ju-Yi Hsieh
1,
*, Yu-Hsiu Chiang
1,
*, Kuan-Yu Chang
1
and Hui-Chih Hung
1,2
1 Department of Life Sciences, National Chung-Hsing University, Taichung, Taiwan
2 Institute of Bioinformatics, National Chung-Hsing University, Taichung, Taiwan
Malic enzyme (ME) comprises a family of oxidative
decarboxylases that catalyze the transformation of the
substrate l-malate to CO
2
and pyruvate, with instanta-
neous reduction of NAD(P)
+
to NAD(P)H [1–3].
Divalent metal ions (Mn
2+
or Mg
2+
) are essential for
this enzymatic reaction. These enzymes are universally
present in nature, with conserved sequences, and have
generally similar structural topology among different

species [4–8]. According to their cofactor specificity,
mammalian ME has been divided into three isoforms:
Keywords
allosteric regulation; analytical
ultracentrifugation; electrostatic interaction;
malic enzyme; mutagenesis
Correspondence
H C. Hung, Department of Life Sciences
and Institute of Bioinformatics, National
Chung-Hsing University, 250, Kuo-Kuang
Road, Taichung, 40227 Taiwan
Fax: +886 4 22851856
Tel: +886 4 22840416 (ext. 615)
E-mail:
*These authors contributed equally to this
work
(Received 30 July 2008, revised
19 November 2008, accepted 4
December 2008)
doi:10.1111/j.1742-4658.2008.06834.x
Here we report on the role of Glu59 in the fumarate-mediated allosteric
regulation of the human mitochondrial NAD(P)
+
-dependent malic enzyme
(m-NAD-ME). In the present study, Glu59 was substituted by Asp, Gln or
Leu. Our kinetic data strongly indicated that the charge properties of this
residue significantly affect the allosteric activation of the enzyme. The
E59L enzyme shows nonallosteric kinetics and the E59Q enzyme displays a
much higher threshold in enzyme activation with elevated activation con-
stants, K

A,Fum
and aK
A,Fum
. The E59D enzyme, although retaining the
allosteric property, is quite different from the wild-type in enzyme activa-
tion. The K
A,Fum
and aK
A,Fum
of E59D are also much greater than those
of the wild-type, indicating that not only the negative charge of this residue
but also the group specificity and side chain interactions are important for
fumarate binding. Analytical ultracentrifugation analysis shows that both
the wild-type and E59Q enzymes exist as a dimer–tetramer equilibrium. In
contrast to the E59Q mutant, the E59D mutant displays predominantly a
dimer form, indicating that the quaternary stability in the dimer interface
is changed by shortening one carbon side chain of Glu59 to Asp59. The
E59L enzyme also shows a dimer–tetramer model similar to that of the
wild-type, but it displays more dimers as well as monomers and polymers.
Malate cooperativity is not significantly notable in the E59 mutant
enzymes, suggesting that the cooperativity might be related to the molecu-
lar geometry of the fumarate-binding site. Glu59 can precisely maintain the
geometric specificity for the substrate cooperativity. According to the
sequence alignment analysis and our experimental data, we suggest that
charge effect and geometric specificity are both critical factors in enzyme
regulation. Glu59 discriminates human m-NAD-ME from mitochondrial
NADP
+
-dependent malic enzyme and cytosolic NADP
+

-dependent malic
enzyme in fumarate activation and malate cooperativity.
Abbreviations
c-NADP-ME, cytosolic NADP
+
-dependent malic enzyme; ES, enzyme–substrate; ME, malic enzyme; m-NAD-ME, mitochondrial NAD(P)
+
-
dependent malic enzyme; m-NADP-ME, mitochondrial NADP
+
-dependent malic enzyme; PFK-1, phosphofructokinase-1.
FEBS Journal 276 (2009) 983–994 ª 2009 The Authors Journal compilation ª 2009 FEBS 983
cytosolic NADP
+
-dependent ME (c-NADP-ME)
[9,10], mitochondrial NADP
+
-dependent ME (m-NADP-
ME) [11], and mitochondrial NAD(P)
+
-dependent ME
(m-NAD-ME) [1,11]. Mitochondrial NAD(P)
+
-depen-
dent ME has dual cofactor specificity, and can use
both NAD
+
and NADP
+
as cofactor, but physiologi-

cally it favors NAD
+
in maximizing its enzyme activ-
ity [1,12]. Human m-NAD-ME may associated with
the growth of highly proliferating tissues and tumors
through the NADH and pyruvate produced in gluta-
minolysis [1,13–20].
Unlike the other two mammalian isoforms,
m-NAD-ME is a regulatory enzyme with a complex
control system for manipulating its catalytic activity
[21–23]. The enzyme exhibits positive cooperative
behavior with respect to the substrate l-malate, and it
is an allosteric enzyme activated by fumarate [18,21–
27]. Previous studies have suggested that ATP may act
as an allosteric inhibitor of m-NAD-ME [18,21,25],
and the allosteric properties of this isoform may relate
to its particular role in the pathways of malate and
glutamine oxidation in tumor mitochondria [17–21,24].
However, further site-directed mutagenesis and kinetics
studies showed that ATP may actually act as an active
site inhibitor, rather than an allosteric inhibitor
[28,29].
The crystal structures of MEs demonstrate that the
enzyme is a homotetramer with a double-dimer qua-
ternary structure. On the basis of structural informa-
tion, MEs are categorized into a new class of
oxidative decarboxylases with a novel backbone struc-
ture [4,6,8,30]. The structures of human m-NAD-ME
with malate ⁄ pyruvate, Mn
2+

⁄ Mg
2+
, NAD
+
, fuma-
rate and transition state analog inhibitors have been
resolved [4,5,31–33]. In the structure of human
m-NAD-ME, besides the active site, there are two
regulatory sites (Fig. 1A). One of them, located at the
tetramer interface and called the exo site, is occupied
by an NAD or ATP molecule. The other is at the
dimer interface, and is occupied by fumarate [32]. The
structures of pigeon c-NADP-ME and Ascaris suum
m-NAD-ME in complex with various ligands have
also been reported [6,7,30]. These structures do not
show the additional exo site, but a separate allosteric
site is found in the A. suum m-NAD-ME at the dimer
interface [26,27]. Figure 1B shows the binding mode
of fumarate at the dimer interface. Structural studies
have revealed that Arg67 and Arg91 are the ligands
for fumarate binding. The side chains of Arg91 and
Arg67 form salt bridges with the carboxylate group of
fumarate (Fig. 1B). Site-directed mutagenesis and
kinetic studies confirmed that Arg67 and Arg91 are
indeed essential for fumarate activation. Both R67S
and R91T mutant enzymes are insensitive towards
fumarate [32]. However, both Arg67 and Arg91 are
conserved among other ME isoforms that are not
activated by fumarate (Fig. 1C). Thus, additional fac-
tors must be involved in governing the activation

mechanism of fumarate in the human and A. suum
m-NAD-ME [27,32].
In our previous work, we delineated the functional
role of Asp102, which is close to the Arg67–fumarate–
Arg91 ion pair network but does not directly interact
with fumarate and is not conserved in other nonallos-
teric MEs (Fig. 1C). We proposed that Asp102 is
important for preserving the electrostatic balance in
the fumarate-binding pocket, which may be a central
factor in the regulatory mechanism of fumarate. Muta-
tion of Asp102 to Ala and Lys, however, abolishes the
allosteric activation of the enzyme [23].
In this article, we aimed to explore in detail the fac-
tors governing the allosteric regulation of the enzyme.
Previous studies have already shown that Glu59, a
structural neighbor of Arg67, plays an important role
in the fumarate activation of the enzyme. Besides the
R67S and R91T mutant enzymes, which are insensitive
towards fumarate, the E59L mutant enzyme does not
show any enzyme activation with fumarate, indicating
that Glu59 has remarkable effects on allosteric activa-
tion [32]. In the present study, the functional roles of
Glu59 in the regulatory mechanism of the enzyme are
elucidated. Glu59 is substituted by Asp, Gln and Leu.
Detailed kinetic and analytical ultracentrifugation
analyses of these mutants help to determine the
factors affecting fumarate activation, subunit–subunit
interaction and substrate cooperativity of human
m-NAD-ME.
Results

Kinetic parameters of the human wild-type and
E59 mutant m-NAD-MEs
The kinetic parameters of the wild-type and E59
mutant enzymes were determined with and without
20 mm fumarate (Table 1). Without fumarate, there
were no significant differences in K
m,NAD
and
K
0.5,Malate
observed among the wild-type and E59
mutant enzymes, except for the E59D enzyme. The
K
m,NAD
and K
0.5,Malate
values of E59D were 2-fold and
4.4-fold higher, respectively, than that of the wild-type.
Furthermore, the k
cat
value of E59 mutant enzymes
was only one-third to one-quarter of that of the
wild-type, suggesting that mutation of Glu59 in the
fumarate-binding site causes the enzyme to become less
efficient in catalysis.
Human mitochondrial NAD(P)
+
-dependent malic enzyme J Y. Hsieh et al.
984 FEBS Journal 276 (2009) 983–994 ª 2009 The Authors Journal compilation ª 2009 FEBS
A

C
D
B
Fig. 1. Fumarate-binding site for human m-NAD-ME. (A) Homotetramer of human m-NAD-ME (Proein Data Bank code: 1PJ3). (B) Fumarate-
binding site of m-NAD-ME. The corresponding amino acids in the fumarate-binding site, Arg67, Arg91 and Glu59, are represented as a ball-
and-stick model. The color is yellow for fumarate. This figure was generated with
PYMOL (DeLano Scientific LLC, San Carlos, CA, USA). (C)
Multiple sequence alignments of three clusters of ME isoforms around the fumarate-binding region are shown. Amino acid sequences of
MEs were obtained by a similarity search of
BLAST [44], and alignments were created with CLUSTAL W [45]. This figure was generated using
the
BIOEDIT sequence alignment editor program [46]. (D) Glu59-binding ligands in the wild-type enzyme are shown as the LIGPLOT diagram
[47]. The bold bonds indicate the specific amino acid, the thin bonds are the hydrogen-bonded residues, and the green dashed lines corre-
spond to the hydrogen bonds. Spoked arcs represent hydrophobic contacts.
J Y. Hsieh et al. Human mitochondrial NAD(P)
+
-dependent malic enzyme
FEBS Journal 276 (2009) 983–994 ª 2009 The Authors Journal compilation ª 2009 FEBS 985
With fumarate, the K
m,NAD
and K
0.5,Malate
of the
wild-type enzyme decreased, the k
cat
values of the
enzyme increased, and the h value, which represents
the cooperativity of malate binding, was significantly
reduced from 1.8 to 1, indicating the characteristics of
allosteric regulation of the enzyme isoform by fuma-

rate. The E59Q enzyme, similar to the wild-type,
showed a decrease in K
m,NAD
and K
0.5,Malate
but an
increase in k
cat
. However, a considerable decrease in
malate cooperativity was observed in this mutant. The
h value was 1.2 with or without fumarate. For the
E59D enzyme, the K
m,NAD
and K
0.5,Malate
values
decreased by about two-fold and 10-fold, respectively,
and the k
cat
value of the mutant enzyme increased by
three-fold. The malate cooperativity, however, com-
pletely disappeared in this mutant. The h value was
1.0, with or without fumarate. Indeed, the E59D
enzyme had an unusually large K
0.5,Malate
value, which
could be reduced to a level similar to that of the wild-
type by the addition of fumarate, suggesting that the
active site of this mutant had been changed and was
readjusted by fumarate. For the E59L enzyme, the

K
0.5,Malate
and k
cat
values were not notably influenced
by fumarate, indicating that these mutant enzymes
were insensitive to fumarate activation.
Activating effect of fumarate on the human
wild-type and E59 mutant m-NAD-MEs
The initial rates of m-NAD-ME measured in various
concentrations of fumarate showed hyperbolic kinetics
(Fig. 2). At a saturating concentration of fumarate, the
maximal activation by fumarate for the wild-type
enzyme was approximately 1.5-fold, with an apparent
K
A
value of 0.21 ± 0.03 mm (Fig. 2, closed circles).
The maximal activation fold of the E59Q enzyme was
similar to that of the wild-type enzyme, whereas the
relative enzyme activity of the E59Q enzyme was only
one-third of that of the wild-type (Fig. 2, closed trian-
gles), and it had a much higher apparent K
A
value
(14.1 ± 2.3 mm) than that of the wild-type, suggesting
that the E59Q enzyme needed more fumarate mole-
cules to achieve its maximal activation. The E59D
enzyme could be activated more than three-fold, with
an apparent K
A

value of 6.1 ± 1.1 mm, a value that
was also much higher than that of the wild-type
(Fig. 2, open circles). Even though the maximal activa-
tion fold of the E59D enzyme was more than that of
the wild-type, the relative enzyme activity of the E59D
enzyme was raised to the level of one-half of that of
the wild-type (Fig. 2, open circles). Like the E59Q
enzyme, the E59D enzyme also required more fuma-
rate to achieve this activation. The E59L enzyme could
not be activated by fumarate (Fig. 2, open triangles).
These preliminary results indicate that the side chain
properties of residue 59 seem to have a great impact
on the allosteric regulation of ME.
Activation constants of the human wild-type and
E59 mutant m-NAD-MEs
The activation constants of the wild-type and E59
mutant enzymes were further determined by kinetic
analysis. The enzyme activities were assayed in a broad
range of substrate malate concentrations at different
fixed concentrations of fumarate (data not shown).
These curves were globally fitted to Eqn (1), and the
activation constant of fumarate for free enzyme
(K
A,Fum
) and for the enzyme–substrate (ES) complex
(aK
A,Fum
) were estimated.
Table 1. Kinetic parameters for the human wild-type and E59
mutant m-NAD-MEs. ), no fumarate added; +, with 20 m

M fuma-
rate added.
K
m,NAD
(mM)
K
0.5,Malate
(mM) k
cat
(s
)1
) h
Wild-type
) 0.88 ± 0.10 15.44 ± 0.97 208 ± 10.6 1.84 ± 0.21
+ 0.43 ± 0.04 3.28 ± 0.37 332 ± 15.3 1.01 ± 0.09
E59Q
) 0.92 ± 0.10 15.23 ± 2.10 74 ± 7.5 1.24 ± 0.13
+ 0.82 ± 0.06 7.01 ± 1.13 141 ± 10.3 1.21 ± 0.07
E59D
) 1.60 ± 0.16 67.22 ± 7.76 57 ± 3.5 1.03 ± 0.05
+ 0.76 ± 0.04 6.43 ± 0.68 196 ± 8.6 1.01 ± 0.11
E59L
) 0.76 ± 0.06 10.68 ± 1.03 53 ± 3.1 1.20 ± 0.14
+ 0.49 ± 0.04 10.75 ± 1.91 50 ± 4.1 1.00 ± 0.08
[Fumarate] (mM)
010203040
Velocity (µM·min
–1
)
0

20
40
60
80
100
Fig. 2. Fumarate activation of the human wild-type and E59 mutant
m-NAD-MEs. The assay mixture contained ME (1.5 lg), 40 m
M
malate, 10 mM MgCl
2
, and 2 mM NAD
+
, with various fumarate
concentrations as indicated. Closed circles, wild-type enzyme;
closed triangles, E59Q enzyme; open circles, E59D enzyme; open
triangles, E59L enzyme.
Human mitochondrial NAD(P)
+
-dependent malic enzyme J Y. Hsieh et al.
986 FEBS Journal 276 (2009) 983–994 ª 2009 The Authors Journal compilation ª 2009 FEBS
The activation parameters of the wild-type and E59
mutant human enzymes are summarized in Table 2.
The values of K
A,Fum
and aK
A,Fum
with respect to
malate for the wild-type enzyme were 0.73 mm and
0.17 mm, respectively, indicating that fumarate binds
the ES complex more tightly than free enzyme. For the

E59Q enzyme, the values of K
A,Fum
and aK
A,Fum
were
13.1 and 5.63 mm, respectively, greater than those of
the wild-type by at least an order of magnitude (18-
fold and 33-fold, respectively), indicating that the bind-
ing affinity of fumarate for the E59Q enzyme was
markedly less than that of the wild-type. The negative
charge of Glu59 is important for the binding affinity
of fumarate to either free enzyme or the ES complex.
For the E59D enzyme, the K
A,Fum
and aK
A,Fum
values
were 8.5 and 3.91 mm, respectively, also larger than
those of the wild-type enzyme by over 10-fold. This
reveals that although the negative charge was con-
served, the activation constants of fumarate for the
E59D enzyme were considerably elevated, reflecting a
decreased binding affinity of fumarate.
Self-association of the human wild-type and E59
mutant m-NAD-MEs
The fumarate-binding site resides at the dimer interface
(Fig. 1A). We use analytical ultracentrifugation to
examine the possible change in the quaternary struc-
ture of E59 mutant enzymes. Figure 3 shows the con-
tinuous sedimentation coefficient distribution of the

wild-type and E59 mutants. The sedimentation coeffi-
cients of 6.5 S and 9.0 S represented the dimer and
tetramer, respectively, corresponding to molecular
masses of 124 and 248 kDa. The quaternary structure
of the wild-type is in dimer–tetramer equilibrium with
different protein concentrations (Fig. 3A–C). The
E59Q enzyme apparently shows a similar dimer–
tetramer pattern as the wild-type (Fig. 3D–F), both
being in dimer form in low protein concentrations and
being reconstituted into tetramers in high protein
concentrations. In contrast to the E59Q enzyme, the
E59D enzyme has a predominantly dimer form, with a
few monomers and polymers in the range of protein
concentrations used (Fig. 3G–I), indicating that the
quaternary stability in the dimer interface is greatly
changed by shortening one carbon side chain of Glu59
to Asp59. The E59L enzyme also showed a dimer–
tetramer model similar to that of the wild-type, but it
displayed more dimers as well as monomers and poly-
mers, which are not observed to a significant extent in
the wild-type WT (Fig. 3J–L).
In order to estimate the self-association of enzymes
quantitatively, the sedimentation velocity data were
analyzed globally to determine the dissociation constant
(K
d
) of the wild-type and E59 mutants (Fig. 4). Struc-
tural data show that the interactions in the tetramer
interface are weaker than those in the dimer interface
(Fig. 1A); thus, the K

d
value of the wild-type may
reflect the dissociation between the A and D or B and C
subunits to form AB or CD dimers. The wild-type and
E59Q enzymes had similar K
d
values of 6.3 and 7.9 lm,
respectively (Table 2), showing that the subunit–subunit
interactions in the dimer interface were not disrupted
by substituting a negatively charged Glu with a
neutrally charged Gln. The E59D enzyme, although
conserving the negative charge on residue 59, still
altered its dimer–tetramer equilibrium into a dominant
dimer form with a K
d
value of 4395 lm, which is over
600-fold larger than that of the wild-type. As Glu59 is
near the dimer interface, the E59D dimer might be an
AD (BC) dimer. If this is the case, the K
d
value of the
E59D enzyme may represent the dissociation in the
dimer interface. The E59L enzyme is also in dimer–
tetramer equilibrium, with a slightly larger K
d
value
than the wild-type, suggesting that the Leu substitution
did not cause substantial changes in the dimer interface.
In the presence of fumarate, the dimer–tetramer
equilibrium of the wild-type was shifted (Fig. 5A).

With the addition of fumarate, most dimeric enzymes
reconstitute into tetrameric enzymes, suggesting that
fumarate stabilizes the tetrameric state of the enzyme,
which may increase the catalytic activity. Similar to
the wild-type, the E59Q enzyme was also changed in
its dimer–tetramer equilibrium by fumarate (Fig. 5B),
suggesting that the tetramer organization was not
perturbed in this mutant. The E59D enzyme, however,
could not be reconstituted into a tetramer by fumarate
(Fig. 5C); it existed in a dimeric form in the presence
of fumarate. Although the quaternary structure of the
E59L enzyme displayed a model similar to that of the
wild-type (Fig. 3), the dimer–tetramer equilibrium of
this mutant enzyme could not be shifted by fumarate
(Fig. 5D), suggesting that the mutant enzyme is insen-
sitive to fumarate with regard not only to catalytic
activity but also to quaternary structure organization.
Table 2. Activation constants of fumarate and dissociation con-
stants of dimer–tetramer equilibrium for the human wild-type and
E59 mutant m-NAD-MEs.
K
A,Fum
(mM)
aK
A,Fum
(mM) K
d
(lM)
DG
(kcalÆmol

)1
)
Wild-type 0.73 ± 0.11 0.17 ± 0.03 6.27 ± 0.05 )7.0 ± 0.05
E59Q 13.1 ± 3.9 5.63 ± 1.68 7.93 ± 0.06 )6.8 ± 0.05
E59D 8.5 ± 1.8 3.91 ± 0.83 4395 ± 23 )3.2 ± 0.02
E59L – – 10.21 ± 0.10 )6.7 ± 0.07
J Y. Hsieh et al. Human mitochondrial NAD(P)
+
-dependent malic enzyme
FEBS Journal 276 (2009) 983–994 ª 2009 The Authors Journal compilation ª 2009 FEBS 987
Discussion
Binding network of the allosteric activator
fumarate in human m-NAD-ME
Fumarate has been identified as the allosteric activator
for human m-NAD-ME by decreasing the K
m
values
of the active site ligands [13,21,23,28]. As well as the
mammal enzymes, m-NAD-ME from A. suum is also
activated by fumarate [27]. As the crystal structure of
human m-NAD-ME in complex with fumarate has
been determined, the binding network of fumarate in
the enzyme has become clear. In the fumarate-binding
pocket, two arginyl residues, Arg67 and Arg91, have
been identified as determining the major binding affin-
ity of fumarate for the enzyme. An anionic amino
Fig. 3. Continuous sedimentation coefficient distribution of the human wild-type and E59 mutant m-NAD-MEs. The enzymes were used at
three protein concentrations, 0.2, 0.6 and 1.2 mgÆmL
)1
in 50 mM Tris ⁄ HCl buffer (pH 7.4) at 20 °C. (A–C) Wild-type. (D–F) E59Q. (G–I) E59D.

(J–L) E59L.
Human mitochondrial NAD(P)
+
-dependent malic enzyme J Y. Hsieh et al.
988 FEBS Journal 276 (2009) 983–994 ª 2009 The Authors Journal compilation ª 2009 FEBS
acid, Glu59, which forms salt bridges with Arg67 in
the structure but is not found in nonallosteric MEs, is
suggested to be important for fumarate activation [32].
In this study, we examined the effect of the charge and
hydrogen bonding network of the Glu59 side chain on
allosteric regulation of the enzyme.
Charge effect of residue 59 on enzyme regulation
In the structure of the wild-type enzyme, Glu59 is ion-
paired and hydrogen-bonded with Arg67 and Lys57
(Fig. 1D). Simultaneously, fumarate is coordinated
with Arg67 and Arg91 in the structure; the ionic pairs
Fig. 4. Global analysis of sedimentation velocity data of the human wild-type and E59 mutant m-NAD-MEs at three protein concentrations.
Sedimentation was performed at 20 °C with an An-50 Ti rotor and at a rotor speed of 42 000 r.p.m. (A–C) Concentrations of the protein
were 0.2, 0.6 and 1.2 mgÆmL
)1
, respectively. The symbols are raw sedimentation data, and the lines are data fitted by the software SEDPHAT.
(D–F) The fitting residuals of the model from the upper panel afford a reliable analysis result for a dissociation constant (K
d
) of the mono-
mer–dimer equilibrium.
Fig. 5. Continuous sedimentation coefficient distribution of the human wild-type and E59 mutant m-NAD-MEs in the presence of fumarate.
The enzyme concentration was fixed at 0.2 mgÆmL
)1
without (solid lines) or with (broken lines) 1 mM (broken lines) or 3 mM (dotted lines)
fumarate. (A) Wild-type. (B) E59Q. (C) E59D. (D) E59L.

J Y. Hsieh et al. Human mitochondrial NAD(P)
+
-dependent malic enzyme
FEBS Journal 276 (2009) 983–994 ª 2009 The Authors Journal compilation ª 2009 FEBS 989
and hydrogen bonding network among Glu59, Arg67,
fumarate and Arg91 may be responsible for fumarate-
induced allosteric regulation (Fig. 1B). Our kinetic
data strongly indicated that the charge properties of
Glu59 significantly affect the allosteric activation of
the enzyme (Fig. 2). The fact that the E59L enzyme
shows nonallosteric kinetics and the E59Q enzyme dis-
plays a much higher threshold for enzyme activation
clearly indicates that the charge effect is an influential
factor in fumarate activation. Abolishing the salt
bridges between Glu59 and Arg67 has a significant
effect on the binding network of fumarate.
The E59Q enzyme can still be activated by fumarate,
and the maximal activation fold is similar to that of
the wild-type, whereas achieving the maximal activa-
tion requires a higher fumarate concentration. Signifi-
cant differences in K
A,Fum
and aK
A,Fum
were observed
for the E59Q enzyme, suggesting that the negative
charge of Glu has a significant impact on the binding
affinity of fumarate. As the molecular dimensions and
polarity of Gln are close to those of Glu, the proper
conformational geometry and hydrogen bonding net-

work should be preserved even though the negative
charge of Glu is replaced by a neutral side chain of
Gln. Thus, the delayed activation of this mutant
enzyme could be attributed to the ion pairs derived
from Glu59 being destroyed by this replacement.
For the E59D enzyme, despite the negative charge
of the residue and the allosteric property of the
enzyme being retained, the kinetic properties are quite
different from those of the wild-type. The K
A,Fum
and
aK
A,Fum
values of the E59D enzyme are also much
greater than those of the wild-type, indicating that not
only the negative charge of this residue but also the
group specificity and side chain interactions are impor-
tant for fumarate binding. Furthermore, kinetic analy-
sis demonstrates that the E59D enzyme resembles a
partial inactive enzyme with an anomalously high K
m
value and low enzyme activity in the absence of fuma-
rate. This may result from shortening the length of the
side chain in the enzyme, causing geometrical changes
in the fumarate-binding pocket, which may have an
effect on the active site. The structural change in the
E59D enzyme may be adjusted by binding of fumarate,
and the mutant enzyme can thus be reactivated.
Subunit–subunit interaction in the dimer
interface and malate cooperativity

Analytical ultracentrifugation analysis demonstrates
that the wild-type enzyme exists in dimer–tetramer
equilibrium in solution. The quaternary structure of
the E59Q enzyme is as stable as that of the wild-type,
demonstrating a similar dimer–tetramer pattern (Figs 3
and 5) and a similar dissociation constant (Table 2).
This fact suggests that the molecular geometry in the
dimer interface is not significantly changed by the
substitution of Gln.
Unlike the E59Q enzyme, the E59D enzyme exists
mainly as a dimer rather than in dimer–tetramer equilib-
rium (Fig. 3). The dissociation of subunits in the dimer
interface might be caused by the structural change in the
E59D enzyme. The ionic interactions between Asp59
and Lys57 might still occur, but be somewhat altered
because of the shorter side chain of Asp59. In the wild-
type, Lys57 is located in the dimer interface; its side
chain is hydrogen-bonded not only with Glu59 but also
with Pro216 and Tyr218 from the other subunit. Analyt-
ical ultracentrifugation analysis shows that mutation of
Lys57 causes the enzyme to dissociate into unstable
dimers and, further, to form polymers (J Y. Hsieh, Y
W. Fang & H C. Hung, unpublished results). In the
E59D enzyme, Lys57 might be pulled by Asp59 into one
subunit, thus being no longer hydrogen-bonded with
Pro216 and Tyr218 from the other subunit, and finally
leading to the disintegration of tetramers of the enzyme.
This can be fitted well with the explanation of why the
E59L enzyme is kept in dimer–tetramer equilibrium. It
can be concluded that the hydrophobic Leu introduced

did not influence the subunit interactions of Lys57.
However, the monomer and polymer appearing in the
equilibrium may have been caused by the environmental
alteration from hydrophilic to hydrophobic.
Although Glu59 is near the dimer interface, it is also
possible that the E59D dimers are AB (and CD)
dimers, based on the fact that the tetramer is a dimer
of AB (and CD) dimers. Another possibility could be
that the E59D enzyme maintains a disrupted AB
dimer, and the conformational disturbance somehow
prevents the formation of the tetramer. However, this
question can be definitively answered until the
biophysical data for discrimination between AB and
AD dimers are available.
The malate cooperativity in the E59 mutant enzymes
is almost abolished. This is not surprising in the E59D
enzyme, because this mutant enzyme is mainly in
dimer form. For the E59Q and E59L enzymes, even
though the former conserves the allosteric activation of
fumarate, and like the wild-type, both principally
reserve a dimer–tetramer equilibrium, their malate
cooperativity still decreases significantly. The loss of
cooperativity might be related to the alteration in
molecular geometry of the fumarate-binding site. Our
data suggest that only Glu59 can precisely hold the
geometric specificity of the allosteric site and that this
is important for substrate cooperativity.
Human mitochondrial NAD(P)
+
-dependent malic enzyme J Y. Hsieh et al.

990 FEBS Journal 276 (2009) 983–994 ª 2009 The Authors Journal compilation ª 2009 FEBS
Involvement of the allosteric triad in the
regulatory mechanism of human m-NAD-ME
Structural data have clearly revealed that both Arg67
and Arg91 are the direct ligands for fumarate. It is not
surprising that mutation of these argininyl residues
causes the enzyme to become insensitive to fumarate
[32]. According to the sequence alignments, neither
Arg67 nor Arg91 are specifically conserved in m-NAD
ME, the only isoform that could be activated by fuma-
rate. In fact, Arg91 is highly conserved among all clas-
ses of MEs, whereas Arg67, although moderately
conserved, is completely conserved among all mam-
malian ME isoforms (Fig. 1E). Glu59, however, is
conserved only in m-NAD-ME (Fig. 1C); in m-NADP-
ME and c-NADP-ME, this residue is replaced by Leu
or Asn, respectively. Hence, the latter two enzymes are
not fumarate-activated. In fact, the E59L and E59Q
enzymes display, as such, the properties of a nonregu-
latory ME. The E59L m-NAD-ME is insensitive to
fumarate, showing nonallosteric and noncooperative
kinetics, whereas E59Q m-NAD-ME displays a much
higher activation threshold and less cooperativity.
Glu59 thus discriminates human m-NAD-ME from
m-NADP-ME and c-NADP-ME in fumarate regula-
tion and malate cooperativity. Glu59, Arg67 and
Arg91 form an allosteric triad in the fumarate site of
human m-NAD-ME. The allosteric triad is the basic
element for fumarate binding, and thus determines
whether the enzyme is allosteric or nonallosteric.

Many enzymes in the metabolic pathway are
controlled by its allosteric regulator
Besides human m-NAD-ME, many enzymes in the
metabolic pathway are allosteric enzymes. The one
that is most well characterized is phosphofructokinase-
1 (PFK-1) [34]. PFK-1 catalyzes the phosphorylation
of fructose 6-phosphate to fructose 1,6-bisphosphate.
Fructose 2,6-bisphosphate is the potent allosteric acti-
vator for this regulatory enzyme [35,36]. In the absence
of fructose 2,6-bisphosphate, the enzyme is almost
inactive at the physiological concentrations of its sub-
strate, fructose 6-phosphate. When fructose 2,6-bis-
phosphate binds to the allosteric site on PFK-1, it
increases the substrate affinity, with a significant
decrease of K
0.5
from 2 mm to 0.08 mm. Similar to the
effect of fumarate on human m-NAD-ME, fructose
2,6-bisphosphate activates PFK-1 by increasing the
apparent affinity for fructose 6-phosphate. The regula-
tory mechanism of these two enzymes gives a good
example of the reactivation–deactivation of an alloste-
ric enzyme controlled by its specific allosteric regulator
produced in the metabolic pathway.
Experimental procedures
Expression and purification of recombinant MEs
The detailed expression and purification protocols for
human m-NAD-ME have been reported in earlier studies
[1,31]. In brief, m-NAD-ME was subcloned into the expres-
sion vector (pRH281) and transformed into Escherichia coli

BL21 cells for enzyme overexpression by controlling the
inducible trp promoter system [1]. Anionic exchange,
DEAE–Sepharose (Amersham Biosciences, Uppsala, Swe-
den), followed by ATP–agarose affinity chromatography
(Sigma, St Louis, MO, USA) were employed in the enzyme
purification. The purified enzyme was subsequently buffer-
exchanged and concentrated in 30 mm Tris ⁄ HCl (pH 7.4)
and 2 mm b-mercaptoethanol by a centrifugal filter device
(Amicon Ultra-15; Millipore, Billerica, MA, USA) with a
molecular mass cutoff of 30 kDa. The enzyme purity was
checked by SDS ⁄ PAGE, and the protein concentrations
were estimated by the Bradford method [37].
Site-directed mutagenesis
Site-directed mutagenesis was carried out using the Quik-
Change kit (Stratagene, La Jolla, CA, USA). The purified
DNA of human m-NAD-ME was used as a template, and
the primers with the desired codons were employed to change
Glu59 into Asp, Gln and Leu, using a high fidelity of Pfu
DNA polymerase in the PCR reaction. Primers including
the mutation site are 25- to 45-mer, which is considered nec-
essary for specific binding of template DNA. The synthetic
oligonucleotides used in these site-directed mutagene-
sis experiments were 5¢-GGACTTCTACCTCCCAAAATA
GACACACAAGATATTCAAGCC-3¢ for E59D, 5¢-GGAC
TTCTACCTCCCAA AATA
CAGACACAAGATA TTCAA
GCC-3¢ for E59Q, and 5¢-GGACTTCTACCTCCCAAAA
TA
CTGACACAAGATATTCAAGCC-3¢ for E59L. The
nucleotides underlined and marked in bold indicate the

mutation positions. After 16–18 temperature cycles,
the mutated plasmids including staggered nicks were made.
The PCR products were subsequently treated with DpnIto
digest the wild-type human m-NAD-ME templates.
Finally, the nicked DNA with desired mutations was trans-
formed into E. coli strain XL-1, and their DNA sequences
were checked by autosequencing.
Enzyme kinetic analysis
Enzymatic activity of MEs was measured by the reduction
of NAD
+
to NADH. The reaction mixture contained
J Y. Hsieh et al. Human mitochondrial NAD(P)
+
-dependent malic enzyme
FEBS Journal 276 (2009) 983–994 ª 2009 The Authors Journal compilation ª 2009 FEBS 991
50 mm Tris ⁄ HCl (pH 7.4), 40 mm malate (pH 7.4), 2.0 mm
NAD
+
and 10 mm MgCl
2
in a total volume of 1 mL. The
absorbance at 340 nm at 30 °C was instantaneously traced
after the enzyme was added to the reaction mixture, and
monitored continuously in a Beckman DU 7500 spectro-
photometer. Under these conditions, 1 unit of the enzyme
was defined as the amount of enzyme catalyzing the produc-
tion of 1 lmol of NADH per min. An extinction coefficient
of 6.22 per mm for NADH was utilized in the calculations.
Apparent Michaelis constants of the substrate and cofactors

were determined by varying the concentration of one sub-
strate (or cofactor) around its K
m
value, while keeping other
components constant at the saturating concentrations. The
nonessential activation model was employed to estimate the
dissociation constants for free enzyme (E) and ES complex
[38]. The experiment was carried out at a series of fumarate
concentrations and at different concentrations of l-malate.
The total set of data was globally fitted to the following
equation, which was derived from a nonessential activation
mechanism (Scheme 1) [38]:
m=V
max
¼½S=fK
s
Ãð1 þ½A=K
A;Fum
Þ=ð1 þ b½A=aK
A;Fum
Þ
þ½SÃð1 þ½A=K
A;Fum
Þ=ð1 þ b½A=aK
A;Fum
Þg
ð1Þ
in which v is the observed initial velocity, and V
max
is the

maximum rate of the unactivated reaction. The maximum
rate in the presence of fumarate is bV
max
. K
s
is the Micha-
elis constant for the substrate, and K
A,Fum
and aK
A,Fum
are
the activation constants for fumarate binding to free
enzyme (E) and ES complex, respectively.
The sigmoidal curves of [malate] versus initial rates
were fitted into the Hill equation, and data were further
analyzed to calculate the K
0.5
value, the substrate con-
centration at half-maximal velocity, and the Hill coeffi-
cient (h), which were employed to assess the degree of
cooperativity.
m ¼ V
max
½malate
h
=ðK
h
0:5
þ½malate
h

Þ
All data-fitting work was carried out with the sigma -
plot 8.0 program (Jandel, San Rafael, CA, USA).
Quaternary structure analysis by analytical
ultracentrifugation
Sedimentation velocity experiments were carried out using
a Beckman Optima XL-A analytical ultracentrifuge. Sample
(380 lL) and buffer (400 lL) solutions were loaded into the
double sector centerpiece separately, and built up in a
Beckman An-50 Ti rotor. Experiments were performed at
20 °C and a rotor speed of 42 000 r.p.m. Protein samples
were monitored by UV absorbance at 280 nm in a continu-
ous mode with a time interval of 480 s and a step size of
0.002 cm. Multiple scans at different time points were fitted
to a continuous size distribution model by the program
sedfit [39–42]. All size distributions were solved at a confi-
dence level of P = 0.95, a best-fitted average anhydrous
frictional ratio (f ⁄ f
0
), and a resolution N of 200 sedimenta-
tion coefficients between 0.1 and 20.0 S.
To precisely determine the dissociation constants of MEs
in dimer–tetramer equilibrium, sedimentation velocity
experiments were performed with three different protein
concentrations of the enzyme. All sedimentation data were
globally fitted to the monomer–dimer equilibrium model of
the program sedphat to calculate the dissociation constant
(K
d
) of the enzyme [41]. The partial specific volume of the

enzyme, solvent density and viscosity were calculated by
the software program sednterp [43].
Acknowledgements
This work was supported by the National Science
Council, ROC (NSC-96-2311-B-005-005 to H C.
Hung), and in part by the Ministry of Education, Tai-
wan, ROC under the ATU plan. We thank Professor
G. G. Chang (Faculty of Life Sciences and the Insti-
tute of Biochemistry, National Yang-Ming University)
for critically reading the manuscript.
References
1 Loeber G, Infante AA, Maurer-Fogy I, Krystek E &
Dworkin MB (1991) Human NAD
+
-dependent
mitochondrial malic enzyme. J Biol Chem 266, 3016–
3021.
2 Cleland WW (2000) Chemical mechanism of malic
enzyme as determined by isotope effects and alternate
substrates. Protein Pept Lett 7, 305–312.
3 Karsten WE, Liu D, Rao GS, Harris BG & Cook PF
(2005) A catalytic triad is responsible for acid–base
chemistry in the Ascaris suum NAD-malic enzyme.
Biochemistry 44, 3626–3635.
4 Xu Y, Bhargava G, Wu H, Loeber G & Tong L (1999)
Crystal structure of human mitochondrial NAD(P)
+
-
dependent malic enzyme: a new class of oxidative
decarboxylases. Structure 7, 877–889.

k
P
K
A,Fum
K
s
ES
E + S
E + P
+
+
A (Fum)
A (Fum)
EA + S
ESA EA + P
K
s

k
P

K
A,Fum
Scheme 1. Nonessential activation mechanism of human mito-
chondrial NAD(P)
+
-dependent malic enzyme.
Human mitochondrial NAD(P)
+
-dependent malic enzyme J Y. Hsieh et al.

992 FEBS Journal 276 (2009) 983–994 ª 2009 The Authors Journal compilation ª 2009 FEBS
5 Yang Z, Floyd DL, Loeber G & Tong L (2000) Structure
of a closed form of human malic enzyme and implications
for catalytic mechanism. Nat Struct Biol 7, 251–257.
6 Yang Z, Zhang H, Hung HC, Kuo CC, Tsai LC, Yuan
HS, Chou WY, Chang GG & Tong L (2002) Structural
studies of the pigeon cytosolic NADP
+
-dependent
malic enzyme. Protein Sci 11, 332–341.
7 Coleman DE, Rao GS, Goldsmith EJ, Cook PF &
Harris BG (2002) Crystal structure of the malic enzyme
from Ascaris suum complexed with nicotinamide
adenine dinucleotide at 2.3 A
˚
resolution. Biochemistry
41, 6928–6938.
8 Chang GG & Tong L (2003) Structure and function of
malic enzymes, a new class of oxidative decarboxylases.
Biochemistry 42, 12721–12733.
9 Loeber G, Dworkin MB, Infante A & Ahorn H (1994)
Characterization of cytosolic malic enzyme in human
tumor cells. FEBS Lett 344, 181–186.
10 Chang GG, Wang JK, Huang TM, Lee HJ, Chou WY &
Meng CL (1991) Purification and characterization of the
cytosolic NADP
+
-dependent malic enzyme from human
breast cancer cell line. Eur J Biochem 202, 681–688.
11 Loeber G, Maurer-Fogy I & Schwendenwein R (1994)

Purification, cDNA cloning, and heterologous expres-
sion of the human mitochondrial NADP
+
-dependent
malic enzyme. Biochem J 304, 687–692.
12 Hsieh JY, Liu GY, Chang GG & Hung HC (2006)
Determinants of the dual cofactor specificity and
substrate cooperativity of the human mitochondrial
NAD(P)+-dependent malic enzyme: functional roles of
glutamine 362. J Biol Chem 281, 23237–23245.
13 Mandella RD & Sauer LAT (1975) The mitochondrial
malic enzymes I. Submitochondrial localization and
purification and properties of the NAD(P)+-dependent
enzyme from adrenal cortex. J Biol Chem 250, 5877–
5884.
14 Sauer LA, Dauchy RT, Nagel WO & Morris HP (1980)
Mitochondrial malic enzymes. Mitochondrial
NAD(P)+-dependent malic enzyme activity and
malate-dependent pyruvate formation are progression-
linked in Morris hepatomas. J Biol Chem 255, 3844–
3848.
15 Baggetto LG (1992) Deviant energetic metabolism of
glycolytic cancer cells. Biochimie 74, 959–974.
16 Sanz N, Diez-Fernandez D, Valverde AM, Lorenzo M,
Benito M & Cascales M (1997) Malic enzyme and glu-
cose 6-phosphate dehydrogenase gene expression
increases in rat liver cirrhogenesis. Br J Cancer 75, 487–
492.
17 Moreadith RW & Lehninger AL (1984) The pathways
of glutamate and glutamine oxidation by tumor cell

mitochondria. Role of mitochondrial NAD(P)+
dependent malic enzyme. J Biol Chem 259, 6215–
6221.
18 Moreadith RW & Lehninger AL (1984) Purification,
kinetic behavior, and regulation of NAD(P)
+
malic
enzyme of tumor mitochondria. J Biol Chem 259, 6222–
6227.
19 Fahien LA & Teller JK (1992) Glutamate–malate
metabolism in liver mitochondria. J Biol Chem 267,
10411–10422.
20 Mckeehan WL (1982) Glycolysis, glutaminolysis, and
cell proliferation. Cell Biol Int Rep 6, 635–650.
21 Sauer LA (1973) An NAD
+
and NAD(P)
+
-dependent
malic enzyme with regulatory properties in rat liver and
adrenal cortex mitochondrial fractions. Biochem Biophys
Res Commun 50, 524–531.
22 Landsperger WJ & Harris BG (1976) NAD
+
-malic
enzyme. Regulatory properties of the enzyme from
Ascaris suum. J Biol Chem 251, 3599–3602.
23 Hung HC, Kuo MW, Chang GG & Liu GY (2005)
Characterization of the functional role of allosteric site
Asp102 in the regulatory mechanism of human mito-

chondrial NAD(P)
+
-dependent malic enzyme. Biochem
J 392, 39–45.
24 Teller JK, Fahien LA & Davis JW (1992) Kinetic and
regulation of hepatoma mitochondrial NAD(P) malic
enzyme. J Biol Chem 267, 10423–10432.
25 Zdnierowicz S, Swierczynski J & Selewski L (1988)
Purification and properties of the NAD(P)
+
-dependent
malic enzyme from human placental mitochondria. Bio-
chem Med Metab Biol 39, 208–216.
26 Lai CJ, Harris BG & Cook PF (1992) Mechanism of
activation of the NAD-malic enzyme from Ascaris suum
by fumarate. Arch Biochem Biophys 299, 214–219.
27 Karsten WE, Pais JE, Rao GS, Harris BG & Cook PF
(2003) Ascaris suum NAD-malic enzyme is activated by
L-malate and fumarate binding to separate allosteric
site. Biochemistry 42, 9712–9721.
28 Hsu WC, Hung HC, Tong L & Chang GG (2004) Dual
functional roles of ATP in the human mitochondrial
malic enzyme. Biochemistry 43, 7382–7390.
29 Hung HC, Chien YC, Hsieh JY, Chang GG & Liu GY
(2005) The functional roles of ATP binding residues in
the catalytic site of human mitochondrial NAD(P)
+
-
dependent malic enzyme. Biochemistry 44, 12737–
12745.

30 Rao GSJ, Coleman DE, Karsten WE, Cook PF & Har-
ris BG (2003) Crystallographic studies on Ascaris suum
NAD-malic enzyme bound to reduced cofactor and
identification of an effector site. J Biol Chem 278,
38051–38058.
31 Bhargava G, Mui S, Pav S, Wu H, Loeber G & Tong L
(1999) Preliminary crystallographic studies of human
mitochondrial NAD(P)
+
-dependent malic enzyme.
J Struct Biol 127, 72–75.
32 Yang Z, Lanks CW & Tong L (2002) Molecular mecha-
nism for the regulation of human mitochondrial
J Y. Hsieh et al. Human mitochondrial NAD(P)
+
-dependent malic enzyme
FEBS Journal 276 (2009) 983–994 ª 2009 The Authors Journal compilation ª 2009 FEBS 993
NAD(P)
+
-dependent malic enzyme by ATP and fuma-
rate. Structure 10, 951–960.
33 Tao X, Yang Z & Tong L (2003) Crystal structures of
substrate complexes of malic enzyme and insights into
the catalytic mechanism. Structure 11, 1141–1150.
34 Hue L & Rider MH (1987) Role of fructose 2,6-bis-
phosphate in the control of glycolysis in mammalian
tissues. Biochem J 245, 313–324.
35 Schirmer T & Evans PR (1990) Structure basis of the
allosteric behavior of phosphofructokinase. Nature 343,
140–145.

36 Hers HG & Van Schaftingen E (1982) Fructose 2,6-bis-
phosphate 2 years after its discovery. Biochem J 206,
1–12.
37 Bradford MM (1976) A rapid and sensitive method for
the quantitation of microgram quantities of protein,
utilizing the principle of protein-dye binding. Anal
Biochem 72, 248–254.
38 Segel IH (1993) Enzyme Kinetics: Behavior and Analysis
of Rapid Equilibrium and Steady-state Enzyme Systems.
Wiley Classics Library, Wiley Interscience, John Wiley
& Sons, New York, NY.
39 Schuck P (2000) Size-distribution analysis of macromol-
ecules by sedimentation velocity ultracentrifugation and
Lamm equation modeling. Biophys J 78, 1606–1619.
40 Lebowitz J, Lewis MS & Schuck P (2002) Modern ana-
lytical ultracentrifugation in protein science: a tutorial
review. Protein Sci 11, 2067–2079.
41 Schuck P (2003) On the analysis of protein self-associa-
tion by sedimentation velocity analytical ultracentrifu-
gation. Anal Biochem 320, 104–124.
42 Schuck P, Perugini MA, Gonzales NR, Howlett GJ &
Schubert D (2002) Size distribution analysis of
proteins by analytical ultracentrifugation: strategies
and application to model systems. Biophys J 82, 1096–
1111.
43 Laue TM, Shah BD, Ridgeway TM & Pelleter SL
(1992) Analytical Ultracentrifugation in Biochemistry and
Polymer Science. The Royal Society of Chemistry,
Cambridge.
44 Altschul SF, Boguski MS, Gish W & Wootton JC

(1994) Issues in searching molecular sequence databases.
Nat Genet 6, 119–129.
45 Higgins D, Thompson J, Gibson T, Thompson JD,
Higgins DG & Gibson TJ (1994) CLUSTAL W:
improving the sensitivity of progressive multiple
sequence alignment through sequence weighting,
position-specific gap penalties, and weight matrix
choice. Nucleic Acids Res 22, 4673–4680.
46 Hall TA (1999) BioEdit: a user-friendly biological
sequence alignment editor and analysis program for
Windows 95 ⁄ 98 ⁄ NT. Nucleic Acids Symp Ser 41,
95–98.
47 Wallace AC, Laskowski RA & Thornton JM (1995)
LIGPLOT: a program to generate schematic diagrams
of protein–ligand interactions. Protein Eng 8, 127–134.
Human mitochondrial NAD(P)
+
-dependent malic enzyme J Y. Hsieh et al.
994 FEBS Journal 276 (2009) 983–994 ª 2009 The Authors Journal compilation ª 2009 FEBS