Recombinant human glucose-6-phosphate dehydrogenase
Evidence for a rapid-equilibrium random-order mechanism
Xiao-Tao Wang
1
, Shannon W. N. Au
1,2
, Veronica M. S. Lam
1,
* and Paul C. Engel
3,
*
1
Department of Biochemistry, The University of Hong Kong, Hong Kong SAR;
2
Section of Structural Biology, Institute of Cancer
Research, Chester Beatty Laboratory, London, UK;
3
Department of Biochemistry and Conway Institute of Biomolecular and
Biomedical Research, University College Dublin, Ireland
Cloning and over-expression of human glucose 6-phosphate
dehydrogenase (Glc6P dehydrogenase) has for the first time
allowed a detailed kinetic study of a preparation that is
genetically homogeneous and in which all the protein mol-
ecules are of identical age. The steady-state kinetics of the
recombinant enzyme, studied by fluorimetric initial-rate
measurements, gave converging linear Lineweaver–Burk
plots as expected for a ternary-complex mechanism. Patterns
of product and dead-end inhibition indicated that the
enzyme can bind NADP
+
and Glc6P separately to form

using enzyme from pooled blood, have variously proposed
either compulsory-order or random-order mechanisms. Our
study appears to provide unambiguous evidence for the
latter pattern of substrate binding.
Keywords: glucose-6-phosphate dehydrogenase; steady-state
kinetics; rapid-equilibrium random-order mechanism;
alternative substrate; product inhibition.
Glucose-6-phosphate dehydrogenase (EC 1.l.1.49) in
humans is an X-chromosome-linked housekeeping enzyme,
vital for the life of every cell. It catalyses the oxidation of
D
-glucose 6-phosphate to
D
-glucono-d-lactone 6-phosphate
in the first committed stepof the pentose phosphate pathway,
which provides cells with pentoses and reducing power in the
form of NADPH. In red blood cells, this is the only source of
NADPH required to protect the cells (via glutathione [1,2]
and catalase [3,4]) against hydrogen peroxide and other
oxidative damage. Accordingly, numerous Glc6P dehydrog-
enase mutations are associated with haemolytic anaemia [5].
Until recently, detailed structural information was avail-
able only for the Glc6P dehydrogenase of Leuconostoc
mesenteroides [6]. Extensive kinetic analysis of the NAD
+
-
and NADP
+
-linked reactions for this bacterial Glc6P
dehydrogenase [7–10] suggests different mechanisms for the

+
as the leading
substrate, whereas Birke et al. [15] obtained quite different
results from similar experiments. Their steady-state kinetic
study, including measurements with inhibitors and alter-
native substrates, suggested a random-order ternary-com-
plex mechanism.
The present study was prompted not only by these
unresolved disagreements but also by the need for a reliable
kinetic description of the normal enzyme as a baseline for
future studies of clinically significant Glc6P dehydrogenase
mutants. Furthermore, our recently solved crystallographic
structure of human Glc6P dehydrogenase [16,17] clearly
vindicates earlier claims that each Glc6P dehydrogenase
subunit has not one but two coenzyme binding sites, and this
in itself demands a careful check of the dependence of
reaction rates on coenzyme concentration. For this reinves-
tigation of the kinetic mechanism, the human Glc6P
Correspondence to P. Engel, Department of Biochemistry,
University College Dublin, Belfield, Dublin 4, Ireland.
Fax: + 353 1283 7211, Tel.: + 353 1716 1547,
E-mail: [email protected]
Abbreviations: Glc6P, glucose 6-phosphate.
*Note: these authors contributed equally to this paper.
(Received 12 February 2002, revised 10 May 2002,
accepted 23 May 2002)
Eur. J. Biochem. 269, 3417–3424 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.03015.x
dehydrogenase gene was cloned and over-expressed in
Escherichia coli so that rate measurements could be made
with freshly prepared, kinetically homogeneous enzyme.

were determined spectrophotometrically at 260 nm (e
260
¼
18.0 · 10
3
M
)1
Æcm
)1
), NADPH at 340 nm (e
340
¼
6.22 · 10
3
M
)1
Æcm
)1
) and deaminoNADP
+
at 249 nm
(e
249
¼ 14.7 · 10
3
M
)1
Æcm
)1
).

minimal medium [0.1
M
KH
2
PO
4
,0.015
M
(NH
4
)
2
SO
4
,
0.8 m
M
MgSO
4
,2l
M
FeSO
4
was adjusted to pH 7.0 with
KOH, and 4 mgÆmL
)1
glucose, 25 lgÆmL
)1
methionine and
histidine were added as supplements] containing

[21] equilibrated with 0.1
M
Tris/HCl buffer, pH 7.6
containing 5% glycerol, 1 m
M
6-amino-n-caproic acid,
3 lgÆmL
)1
aprotinin and 0.1% 2-mercaptoethanol. The
enzyme was eluted with 80 l
M
NADP
+
in this same buffer
and assessed according to the WHO guidelines [22] (data not
shown). Purity was verified by 10% SDS/PAGE [23].
Calibration of the fluorescence emitted by NADPH
and deaminoNADPH
With NADP
+
or deaminoNADP
+
as coenzyme, the activity
of Glc6P dehydrogenase was followed via the increasing
fluorescence of the reduced coenzyme. The measured fluor-
escence change has to be related to the fluorescence of a
known concentration of NADPH or deaminoNADPH.
Because deaminoNADPH of high purity is not available
commercially, the kinetic calibration method of Engel &
Hornby was used, relying on enzymatic production of known

catalysed by Glc6P dehydrogenase, in the nomenclature of
Dalziel [25], is of the form:
e
m
¼ U
o
þ
U
X
½X
þ
U
Y
½Y
þ
U
XY
½X½Y
ð1Þ
where X and Y are sugar phosphate and coenzyme,
respectively. The four / parameters are obtained from
initial-rate measurements at varying concentrations of X for
a series of fixed concentrations of Y. Rearrangement of the
equation shows that the intercepts of primary double
3418 X T. Wang et al. (Eur. J. Biochem. 269) Ó FEBS 2002
reciprocal plots with l/[X] as the variable, for example, are
given by /
0
+ /
Y

M
. A similar experiment was carried out by varying
the Glc6P concentrations from 15 l
M
to 150 l
M
and the
NADPH concentrations again from 0 l
M
to 20 l
M
while
fixing the NADP
+
concentration at 10 l
M
. In analogous
fashion, glucosamine 6-phosphate was used as an inhibitor,
covering the same combinations and ranges of substrate
concentration as used in the experiments with NADPH.
Fluorescence titration studies
Additions of NADP
+
partially quenched the fluorescence
at 345 nm emitted when purified Glc6P dehydrogenase was
excited at 290 nm. If F
E
and F
EL
are the relative fluores-

.
RESULTS
Enzyme preparation
Chromatography on 2¢5¢-ADP Sepharose 4B yielded
recombinant human Glc6P dehydrogenase of 99% purity
or better as judged by SDS/PAGE (data not shown). The
specific activity was about 100 UÆmg
)1
protein. Typically
about 5 mg of purified enzyme could be obtained from 1 L
of E. coli culture. This enzyme behaved identically to
Glc6P dehydrogenase from human cells and showed
identical mobility in native gel electrophoresis (data not
shown). This agrees with the finding of Bautista et al. [26]
that recombinant human Glc6P dehydrogenase expressed
in E. coli behaves similarly to the authentic enzyme from red
cells.
Initial velocity experiments
The strictly linear and converging double reciprocal plots
obtained with different combinations of Glc6P and
NADP
+
(Fig. 1) are consistent with a sequential mechan-
ism, in which both substrates must bind to the enzyme
simultaneously before product formation can occur [25,27].
Fig. 1. Graphs to determine the various / parameters for the reaction
catalysed by human Glc6P dehydrogenase with Glc6P and NADP
+
as
substrates. (A) Primary plots of e/v vs. 1/[NADP

the coenzyme analogue, for which the factor of increase in
individual / constants is at most fourfold to fivefold
(/
Glc6P
). At lower concentrations of sugar phosphate, the
contrast between the natural substrate, Glc6P,andthe
deoxy analogue is greatly accentuated, and this is reflected
in the very high values of /
deoxyGlc6P
and /
NADP
+
deoxyGlc6P
,
which are more than 200-fold larger than the corresponding
parameters for Glc6P (Tables 1–3).
Kinetics of inhibition by NADPH and glucosamine
6-phosphate
Product inhibition patterns also offer useful evidence
regarding enzyme reaction mechanism. In this case,
6-phosphogluconolactone is labile and cannot be obtained
at high enough purity for kinetic experiments. However, it is
possible to determine the effects of NADPH on this reaction
(Figs 3 and 4). The intersection on the vertical axis in
Fig. 3A indicates competitive inhibition with respect to
NADP
+
. The linear secondary plot of the apparent K
m
vs.

M
gave a general noncompetitive (mixed)
inhibition pattern with respect to Glc6P (Fig. 4).
Glucosamine 6-phosphate, chosen as a dead-end inhib-
itor, was found to be competitive with respect to Glc6P
(data not shown) but general noncompetitive (mixed) with
respect to NADP
+
(Fig. 5). The apparent K
m
for Glc6P
obtained here is 50.5 l
M
, similar to 54.8 l
M
calculated from
the Dalziel parameters in Tables 1–3. The K
i
determined for
glucosamine 6-phosphate under these conditions was
1.08 m
M
.
Measurement of dissociation constant of NADP
+
Figure 6 shows that NADP
+
quenches the intrinsic fluor-
escence of Glc6P dehydrogenase. The data are consistent
with a simple binding process with a dissociation constant of

(l
M
2
Æs)
/
NADP
+
Glc6P
/
/
NADP
+
(l
M
)
/
NADP
+
Glc6P
/
/
Glc6P
(l
M
)
/
NADP
+
Glc6P
/

/Glc6P
(l
M
Æs)
/
deamino-
NADP
+
Glc6P
(l
M
2
Æs)
/
deaminoNADP
+
Glc6P
/
/
deaminoNADP
+
(l
M
)
/
deamino-
NADP
+
Glc6P
/

no.
/
o
(s)
/
NADP
+
(l
M
Æs)
/
deoxyGlc6P
(l
M
Æs)
/
NADP
+
deoxy
Glc6P
(l
M
2
Æs)
/
NADP
+
deoxyGlc6P
/
/

(s
)1
)
1 0.031 ± 0.003 0.27 ± 0.01 68 ± 2.5 452 ± 14 1674 6.65 25 32.3
2 0.031 ± 0.001 0.28 ± 0.005 68 ± 1.9 451 ± 8.2 1611 6.63 24 32.3
3 0.031 0.28 68 451 1643 6.64 25 32.3
3420 X T. Wang et al. (Eur. J. Biochem. 269) Ó FEBS 2002
titration appears to reflect the behaviour of only one type of
NADP
+
binding site. In contrast to the effect of coenzyme,
Glc6P produced negligible quenching.
DISCUSSION
Human Glc6P dehydrogenase has K
m
values in the
micromolar concentration range for both the sugar phos-
phate substrate and the coenzyme, and therefore their
reliable estimation requires rate measurements with very
low concentrations of each. The fluorimetric method
employed in this study allowed precise and reproducible
initial-rate measurements even for these low concentrations,
permitting a full analysis of all the initial-rate parameters
[25]. The primary plots were linear over wide ranges of
concentrations for both substrate and coenzyme. The
discovery that there are two NADP
+
binding sites on the
enzyme [17,30,31] had raised the possibility that at low
coenzyme concentrations both sites might contribute to the

Y
and /
XY
[25,29]. There is a correspond-
ing set of relationships if Y is the leading substrate. These
relationships were tested for Glc6P dehydrogenase by
using the alternative substrates deaminoNADP
+
and
deoxyGlc6P (Tables 1–3). Tables 1 and 3 show the results
obtained from the use of alternative sugar phosphate
substrates with the same coenzyme, NADP
+
.IfNADP
+
is the leading substrate, then /
NADP
+
Glc6P
//
Glc6P
should
be equal to /
NADP
+
deoxyGlc6P
//
deoxyGlc6P
. The mean values
obtained, 6.8 l

]ateight
fixed concentrations of deoxyGlc6P. (B) Secondary plots of slopes of
primary plots vs. 1/[deoxyGlc6P]. (C) Secondary plots of intercepts of
primary plots vs. 1/[deoxyGlc6P].
Fig. 3. Product inhibition of the Glc6P dehydrogenase reaction by
NADPH: varied [NADP
+
]. (A) Lineweaver–Burk plots at a fixed
concentration of 60 l
M
Glc6P and varying concentrations of NADP
+
in the presence of a range of NADPH concentrations as indicated. (B)
Secondary replot of K
mapp
vs. the concentrations of NADPH.
Ó FEBS 2002 Human glucose-6-phosphate dehydrogenase mechanism (Eur. J. Biochem. 269) 3421
values for the two sugar phosphate substrates (161 s
)1
vs.
25 s
)1
). A compulsory-order mechanism with NADP
+
as
the leading substrate would therefore appear to be ruled
out.
Tests for a mechanism with the sugar phosphate as the
leading substrate can similarly be made by comparing the
data for alternative coenzymes (Tables 1 and 2). The mean

M
Æs). Correspond-
ingly, the value of /
NADP
+
Glc6P
//
NADP
+
/
Glc6P
(161 s
)1
)is
quite different from the value of /
deaminoNADP
+
Glc6P
/
/
deaminoNADP
+
/
Glc6P
(42 s
)1
). It thus seems that a compul-
sory-order mechanism with Glc6P as leading substrate is also
very unlikely, leaving a rapid-equilibrium random-order
sequential mechanism (Scheme 1) as the remaining option.

tion of substrate Y is a complex function [31] of the
concentration of substrate X:
½B¼
b
0
þ b
1
½Xþb
2
½X
2
a
2
½X
2
þ a
1
½X
ð4Þ
Owing to the higher-order dependence on substrate
concentration, the Lineweaver–Burk plot would not be
strictly linear. Because linear plots were observed in all the
experiments described here without any indication of a
systematic departure from this pattern, this in itself suggests
that the reaction catalysed by human Glc6P dehydrogenase
may involve a rapid-equilibrium rather than a steady-state
random-order mechanism, although admittedly the curva-
Fig. 6. Determination of dissociation constant of NADP
+
by fluores-

respectively [32]. If this is indeed the mechanism of human
Glc6P dehydrogenase, then independent estimates of the
dissociation constant for NADP
+
using Glc6P or deo-
xyGlc6P as substrates (Tables 1 and 3) should be equal
(Table 4). Similarly, the dissociation constant for Glc6P
(Tables 1 and 2) should also be the same regardless of
the coenzyme. As mentioned earlier, the value of
/
NADP
+
Glc6P
//
Glc6P
is 6.8 l
M
, which is almost the same as
/
NADP
+
deoxyGlc6P
//
deoxyGlc6P
(6.6 l
M
). This figure is also
very similar to the value of 8.0 l
M
, independently obtained

competitive with respect to Glc6P and general noncom-
petitive with NADP
+
(Fig. 5) indicating that this
inhibitor can bind to both the free enzyme and the
enzyme–NADP
+
binary complex. Because glucosamine
6-phosphate is an analogue of Glc6P, it seems likely that
Glc6P alsocanbindtobothfreeenzymeandenzyme-
NADP
+
complex. The inhibition studies therefore suggest
that both substrate and coenzyme can bind to the free
enzyme. This in itself points towards a random-order
mechanism, further substantiating the quantitative analy-
sis above.
In summary, this study offers the first clear documenta-
tion of a rapid-equilibrium random-order mechanism for
normal human Glc6P dehydrogenase. The direct demon-
stration of crystal complexes of Glc6P dehydrogenase–
Glc6P and Glc6P dehydrogenase–NADP
+
also tends to
support this conclusion (S. W. N. Au, S. Gover & M. J.
Adams, unpublished data). The discrepancy between the
mechanisms deduced in this present study and in some
previous reports could be due to the heterogeneous origin of
the Glc6P dehydrogenase used in earlier work.
ACKNOWLEDGEMENTS

7. Olive, C., Geroch, M.E. & Levy, H.R. (1971) Glucose 6-phos-
phate dehydrogenase from Leuconostoc mesenteroides. Kinetic
studies. J. Biol. Chem. 246, 2047–2057.
8. Levy, H.R., Ingulli, J. & Afolayan, A. (1977) Identification of
essential arginine residues in glucose-6-phosphate dehydrogenase
from Leuconostoc mesenteroides. J. Biol. Chem. 252, 3745–3751.
9. Haghighi, B., Flynn, T.G. & Levy, H.R. (1982) Glucose 6-phos-
phate dehydrogenase from Leuconostoc mesenteroides. Isolation
and sequence of a peptide containing an essential lysine. Bio-
chemistry 21, 6415–6420.
10. Levy,H.R.,Christoff,M.,Ingulli,J.&Ho,E.M.L.(1983)Glu-
cose-6-phosphate dehydrogenase from Leuconostoc mesenteroides.
Arch. Biochem. Biophys. 222, 473–488.
11. Soldin, S.J. & Balinsky, D. (1968) The kinetic properties of human
erythrocyte glucose- 6-phosphate dehydrogenase. Biochemistry 7,
1077–1082.
12. Schroeder, L., Christoff, M. & Levy, H.R. (1984) Glucose-
6-phosphate dehydrogenase from rabbit erythrocytes. Biochim.
Biophys. Acta 784, 48–52.
13. Rosemeyer, M.A. (1987) The biochemistry of glucose-6-phosphate
dehydrogenase, 6-phosphogluconate dehydrogenase and gluta-
thione reductase. Cell Biochem. Funct. 5, 79–95.
14. Adediran, S.A. (1991) Kinetics of human glucose-6-phosphate
dehydrogenase. Biochimie 73, 1222–1218.
15. Birke, S., Kim, H.W., Periclou, A., Schorsch, B., Grouse, D. &
Craney, C. (1989) Kinetics of human erythrocyte glucose-
6-phosphate dehydrogenase dimers. Biochim. Biophys. Acta 999,
243–247.
16. Au, S.W.N., Naylor, C.E., Gover, S., Vandeputte-Rutten, L.,
Scopes, D.A., Mason, P.J., Luzzatto, L., Lam, V.M.S. & Adams,

Rapid equilibrium random /
AB
//
A
/
AB
//
B
Ping Pong /
B
/
A
Ó FEBS 2002 Human glucose-6-phosphate dehydrogenase mechanism (Eur. J. Biochem. 269) 3423
19. Sambrook, J., Fritsch, E.F. & Maniatis, T. (1989) Molecular
Cloning: a Laboratory Manual, Cold Spring Harbor Laboratory
Press, Cold Spring Harbor, New York.
20. Persico, M.G., Viglietto, G., Martini, G., Toniolo, D., Paonessa,
G., Moscatelli, C., Dono, R., Vulliamy, T., Luzzatto, L. & D’Urso,
M. (1986) Isolation of human glucose 6-phosphate dehydrogenase
(G6PD) cDNA clones: primary structure of the protein and
unusual 5¢ non-coding region. Nucl Acids Res. 14, 2511–2522.
21. De Flora, A., Morelli, A., Benatti, U. & Giuliano, F. (1975) An
improved procedure for rapid isolation of glucose 6-phosphate
dehydrogenase from human erythrocytes. Arch. Biochem. Biophys.
169, 362–363.
22. WHO Scientific Group (1967) Standardization of procedures for
the study of glucose-6-phosphate dehydrogenase. WHO. Tech.
Rep. Series 366, 1–53.
23. Laemmli, U.K. (1970) Cleavage of structural proteins during the
assembly of the head of bacteriophage T4. Nature 227, 680–685.

32. Engel, P.C. (1981) Enzyme Kinetics: the Steady-State Approach,
2nd edn. Chapman & Hall, London.
3424 X T. Wang et al. (Eur. J. Biochem. 269) Ó FEBS 2002