Probing the determinants of coenzyme specificity in
Peptostreptococcus asaccharolyticus glutamate
dehydrogenase by site-directed mutagenesis
John B. Carrigan and Paul C. Engel
School of Biomolecular and Biomedical Science, Conway Institute, University College Dublin, Ireland
An impressive phenomenon in enzymology is the
subtle discrimination that nicotinamide nucleotide-
dependent enzymes can make between NADP(H) and
NAD(H) because the only difference between these
two dinucleotide molecules is a phosphate group esteri-
fied at the 2¢-OH position of the adenosine ribose [1].
Understanding how enzymes achieve this selectivity is
not only of intrinsic scientific interest, but also has
practical application. Re-engineering coenzyme speci-
ficity is a significant goal, not only simply to minimize
coenzyme cost, but also to achieve coenzyme compati-
bility in order to couple two enzyme reactions. Predict-
ably, the 2¢-phosphate or 2¢-OH interaction site of the
enzyme:coenzyme complex has been the focus of initial
Keywords
coenzyme specificity; glutamate
dehydrogenase; NAD(P)
+
; nicotinamide
nucleotides; site-directed mutagenesis
Correspondence
P. C. Engel, School of Biomolecular and
Biomedical Science, Conway Institute,
University College Dublin, Belfield,
Dublin 4, Ireland
Fax: +353 1283 7211

cate that improved affinity for the extra phosphate of NADPH is the pre-
dominant reason for the increased catalytic efficiency with this coenzyme.
The marked difference between the results of replacing E243 with aspartate
and with positive residues implies that the mode of NADPH binding in
naturally occurring NADPH-dependent glutamate dehydrogenases differs
from that adopted in E243K or E243D and in other dehydrogenases.
Abbreviation
GDH,
L-glutamate dehydrogenase.
FEBS Journal 274 (2007) 5167–5174 ª 2007 The Authors Journal compilation ª 2007 FEBS 5167
redesign efforts, and changes have been made on the
basis of 3D comparisons and⁄ or sequence alignments
with similar coenzyme-binding structures that selec-
tively bind the alternative coenzyme. Frequently, the
side chain of a glutamate or aspartate binds the 2¢-OH
group and discriminates against NADP
+
or NADPH.
This important residue was highlighted by Wierenga
and Hol [2], together with a glycine-rich motif, in a
sequence fingerprint that determined coenzyme speci-
ficity in the widespread Rossmann fold [3] of dehydro-
genases. This key acidic residue has been termed the
P7 residue by Baker et al. [4]. Enzymes that exclusively
use NADP(H) usually have a smaller, uncharged resi-
due at this position and positively charged residues
nearby, allowing for better interaction with the
2¢-adenosine phosphate [5]. Protein engineering method-
ologies have provided an opportunity to explore the
generality of such specificity rules with a number of

Figure 1 shows a clear example of the interaction of a
P7 glutamate residue with the adenosine ribose of
NADH, in this case bound to dual-specificity bovine
glutamate dehydrogenase [13]. An exception to this
general trend, however, is the NAD(H)-specific GDH
from Clostridium symbiosum, which has glycine at the
P7 position [3,9,12].
Another NAD(H)-specific GDH in Teller’s align-
ment is from Peptostreptococcus asaccharolyticus, also
a mesophilic, gram-positive anaerobic bacterium, in
which GDH serves the same metabolic role as in
C. symbiosum, catalysing the first step in the unusual
hydroxyglutarate pathway of glutamate fermentation
[14]. Despite this close physiological parallel, the two
GDHs show less than 40% sequence identity. The
P. asaccharolyticus GDH, which has been purified by a
number of groups [15–17], has recently been character-
ized in detail in our laboratory following over-expres-
sion in E. coli [18]. Like the clostridial GDH, this is an
enzyme quite highly specific for NAD(H), with k
cat
⁄ K
m
being approximately 1000-fold greater for NADH than
for NADPH, but it conforms to the more general pat-
tern of a glutamate residue at the P7 position. This
implies that the two enzymes distinguish the cofactors
in different ways [3].
In the present study, we have investigated, by means
of site-directed mutagenesis and steady-state kinetics,

90 mg pure proteinÆL
)1
culture. Both the high yield
of soluble protein and the fact that all the proteins
behaved similarly during ion exchange chromato-
graphy suggest that there was no significant over-
all perturbation of structure resulting from the
mutations.
Coenzyme discrimination in the wild-type
enzyme
As a baseline for these studies, it was necessary to
establish the extent of discrimination between the
two natural nicotinamide cofactors in the unmutated
enzyme. Table 1 shows the separate values of k
cat
and
K
m
and the derived value for the catalytic efficiency,
k
cat
⁄ K
m
. The latter parameter was used to establish
discrimination. Thus, the figures of 7.11 s
)1
Ælm
)1
for
NADH and 6.1 · 10

m
for NADPH, it also demonstrates a 50% increase
in k
cat
.
Table 1. Coenzyme specificity of wild-type and mutant Peptostreptococcus asaccharolyticus glutamate dehydrogenases. Initial rates were
measured with NADH and NADPH in 100 m
M potassium phosphate buffer at pH 7 with 20 mM oxoglutarate and 100 mM ammonium chlo-
ride as the fixed concentrations of substrates. Concentrations of NADH and NADPH were varied to obtain values of k
cat
(column 2) and K
m
(column 3) for both coenzymes under these conditions. Catalytic efficiency (k
cat
⁄ K
m
) values (column 4) are used as the basis for comparing
each of the mutants with wild-type GDH and also, in each case, for calculating the discrimination factor between the two coenzymes. Thus,
column 5 gives the ratio of catalytic efficiency for each mutant enzyme to the corresponding value for the wild-type enzyme with the same
coenzyme. This ratio is the ‘factor change’ brought about by the mutation (e.g. for W244S with NADH 1.43 ⁄ 7.11 ¼ 0.201). Column 6 shows
another ratio obtained from the catalytic efficiencies in column 4, namely the ratio, for each enzyme, of the catalytic efficiency with NADH
to that with NADPH (i.e. the discrimination factor defining the degree of specificity for NADH). This value (i.e. 1170) for the unmutated
enzyme is decreased in all the mutants. Column 7 indicates how many folds this discrimination factor is changed in each case. The final
column shows, for each mutant, the factor change (e.g. 1170-fold discrimination in the wild-type GDH decreases to 130-fold in E243D,
a change by a factor of 8.96).
k
cat
(s
)1
) K

NADPH 3.0 ± 0.482 431 ± 88 6.96 · 10
)3
1.14 205 5.68
E243D
NADH 16.9 ± 1.33 12.3 ± 3 1.38 0.194
NADPH 5.06 ± 2.13 476 ± 100 1.06 · 10
)2
1.73 130 8.96
E243R
NADH 1.71 ± 0.08 38.5 ± 5.5 0.044 6.19 · 10
)3
NADPH 1.67 ± 0.17 38.4 ± 7 0.043 7.05 1.02 1140
E243K
NADH 10.6 ± 0.19 36 ± 1.71 0.294 0.041
NADPH 14.8 ± 0.63 53 ± 5 0.28 46 1.05 1110
D245K
NADH 16.4 ± 0.16 18 ± 0.99 0.911 0.128
NADPH 6.0 ± 0.44 238 ± 3.4 0.025 4.1 36.4 32
J. B. Carrigan and P. C. Engel Changing glutamate dehydrogenase coenzyme specificity
FEBS Journal 274 (2007) 5167–5174 ª 2007 The Authors Journal compilation ª 2007 FEBS 5169
Table 1 also shows for each mutant enzyme a direct
comparison (‘factor change’) of its catalytic efficiency
with NADPH with the corresponding figure for the
unmutated enzyme. E243K, with a 46.0-fold increase
in efficiency compared to wild-type, is by far the best
mutant in this respect. Another residue substitution at
the same site, E243R, allowed a less dramatic 7.5-fold
increase in efficiency, followed by D245K with a 4.1-
fold increase and E243D with a 1.73-fold increase. A
less definite change was observed with W244S, where

combination of improved efficiency with NADPH and
diminished efficiency with NADH. This is much
greater than the shift in discrimination values for
D245K, E243D and W244S, which range from 32- to
8.96- to 5.68-fold, respectively. The replacement of glu-
tamate at the P7 position 243 by the more positively
charged lysine and arginine thus gave the most success-
ful results in our attempts to alter coenzyme specificity.
E243K has a k
cat
⁄ K
m
of 0.28 s
)1
Ælm
)1
with NADPH,
6.3-fold greater than the value for E243R. Both these
mutant enzymes, however, have almost equal activity
with the two reduced coenzymes, thus displaying dual
specificity. Indeed, the kinetic constants obtained for
E243R with the two coenzymes are so strikingly close
that it raises at least the possibility of a change in the
rate-limiting step to a coenzyme-independent process
in the mechanism.
The replacement at the same position with Asp pro-
duced a much more modest shift in specificity, as seen
above, and, even though Asp is widely found at this
position in naturally NADP
+

Æmin
)1
) of wild-type and
mutant enzymes with 0.1 m
M NADPH, 20 mM oxoglutarate and
100 m
M ammonium chloride at different pH values.
pH 6 pH 6.5 pH 7.0 pH 7.5 pH 8
Wild-type 1.32 1.23 0.73 0.32 0.33
E243K 3.27 9.01 17.6 25.3 20.1
E243R 1.83 2.26 3.05 3.41 4.1
D245K 1.12 2.16 1.53 0.87 1.14
E243D 4.93 5.0 2.98 1.47 1.47
W244S 0.29 0.51 0.33 0.34 0.5
Table 3. Specific activity values (lmolÆmg
)1
Æmin
)1
) of wild-type and
mutant enzymes with 0.1 m
M NADH, 20 mM oxoglutarate and
100 m
M ammonium chloride at different pH values.
pH 6 pH 6.5 pH 7.0 pH 7.5 pH 8
Wild-type 9.8 20.6 50 155 174
E243K 5.2 9.6 19 56 54
E243R 0.9 2.3 3 12 11
D245K 5.8 12.5 30 68 101.8
E243D 7.2 15 34 90 102
W244S 3.82 6.78 20 56.5 62

as a direct indication of the effect of the mutations
on the enzyme’s binding affinity for reduced coen-
zyme. Inner filter effects made it impossible to obtain
reliable data with coenzyme concentrations greater
than 70 lm and, accordingly, the higher K
d
values
could not be estimated. K
d
values for the reduced
coenzymes were generally much smaller in the pres-
ence of 0.5 mm oxoglutarate than without and,
indeed, without the other substrate, were generally
too high to measure by this method. In most cases
where a K
d
value could be estimated both with and
without oxoglutarate, the addition of the second sub-
strate tightened coenzyme binding by a factor of at
least 20. The exception to this was D245K, which
yielded a relatively low K
d
value (8.6 lm) for NADH
even without oxoglutarate. However, in this latter
case, the interaction of enzyme and coenzyme caused
a very small change in fluorescence.
The measurements with NADH in the absence of
oxoglutarate show a weakening of binding to varying
extents in every one of the mutants. However, with
NADPH on its own K

of 3.5 lm. D245K and
E243R, which are more efficient than E243D with
NADPH, have much higher K
d
values of 139 lm and
20.4 lm. This is a striking illustration that tight bind-
ing is not necessarily catalytically productive.
The strategy behind creating all of these mutants
was to try and achieve tighter binding of NADP(H). It
can clearly be observed not only that residue 243 (the
P7 residue) is a critical residue in the binding of the
coenzyme, but also that those residues in its immediate
vicinity are of importance. The study of GDH
sequence alignments [12] might have suggested that
replacement of glutamate with aspartate at this posi-
tion would be the best choice. However, the most suc-
cessful strategy was to replace Glu243 with a Lys or
Arg, stabilizing the adenosine phosphate in the way
seen in many other dehydrogenases, and this stabiliza-
tion was also achieved, to a lesser extent, by replacing
the Asp at 245 with a positively charged amino acid.
The fluorescence titrations showed that even with these
mutations binding of NADPH was still too weak to
allow determination of a K
d
value in the absence of
the other substrate. In the presence of oxoglutarate,
however, coenzyme binding was tighter and therefore
measurable and, in particular, gave measurable K
d

W244S 56.5 ± 10 2.3 ± 0.5 –
J. B. Carrigan and P. C. Engel Changing glutamate dehydrogenase coenzyme specificity
FEBS Journal 274 (2007) 5167–5174 ª 2007 The Authors Journal compilation ª 2007 FEBS 5171
fact that these two mutants also gave the highest K
m
values with NADH.
Of the different mutants, E243K is the most success-
ful; even though E243R shows a similar shift in dis-
crimination in the intended direction, its k
cat
value
with NADPH is decreased relative to the wild-type
enzyme, whereas the corresponding figure for E243K
(14.8 s
)1
) is not only increased, but is also close to
50% of the k
cat
of the wild-type enzyme with its natu-
ral coenzyme, NADH. Lysine is closer in size to the
glutamate it replaces than arginine, and it is possible
that the latter is too large to allow optimal orientation
of the coenzyme. Although the binding of NADPH is
actually tighter to E243R than E243K in the presence
of 2-oxoglutarate (Table 4), tighter binding is not
necessarily productive binding.
To date, high-resolution crystallographic data for
P. asaccharolyticus GDH have been elusive [19]. How-
ever, even though direct structural studies of coenzyme
complexes with these mutants would doubtless help to

sodium salt), 2-oxoglutarate (monosodium salt) and
Q-Sepharose were purchased from Sigma (Poole, UK).
Restriction enzymes and T4 DNA ligase were obtained
from New England Biolabs (Ipswich, MA, USA) and Pfu-
turbo DNA polymerase were obtained from Stratagene.
Oligonucleotide primers were obtained from Sigma-Genosys
(Poole, UK).
Expression and purification of the wild-type and
mutated P. asaccharolyticus GDH
The ptac85 plasmid [12,22], which allows genes to be
inserted downstream of the isopropyl thio-b-d-galactoside-
inducible tac promoter, was used for the over-expression of
wild-type and mutated GDH genes in E. coli TG1. PCR
overlap extension and whole plasmid synthesis were used to
generate point mutations. Transformation of the E. coli
TG1 host, growth, induction, harvesting, breakage and
enzyme purification were as described by Carrigan et al.
[18]. This involved utilizing the thermostability of the
enzyme by heating the preparation to 70 °C before binding
to an ion exchange column. SDS ⁄ PAGE gels were used to
check that the over-expressed protein was soluble.
Determination of kinetic parameters
Apparent values of k
cat
and K
m
for the coenzyme, either
NADH or NADPH, at fixed concentrations of the other
two substrates (20 mm 2-oxoglutarate and 100 mm ammo-
nium chloride), were determined for the reductive amina-

wild-type GDH with NADPH), the plentiful supply of pure
enzyme meant that large amounts could be added to pro-
duce a high enough rate for accurate measurement.
Changing glutamate dehydrogenase coenzyme specificity J. B. Carrigan and P. C. Engel
5172 FEBS Journal 274 (2007) 5167–5174 ª 2007 The Authors Journal compilation ª 2007 FEBS
Measurement of K
d
Fluorescence titration was carried out on a Perkin-Elmer
fluorimeter (Perkin-Elmer Life Sciences, Boston, MA,
USA) with excitation set at 290 nm and emission measured
at 400 nm. The protein emission peak is actually at
340 nm, but light absorption by the reduced coenzyme at
this wavelength could cause experimental error. Protein
(20 lg) was added to varying concentrations of NAD(P)H
in 0.1 m potassium phosphate buffer at pH 7 and at a con-
stant temperature of 25 °C. An attempt was made to obtain
values in the presence of 0.5 mm oxoglutarate as well as
without. The changes in fluorescence of the protein were
plotted versus coenzyme concentration. The data were fitted
to a saturation plot using enzpack which provided an esti-
mate of the K
d
.
Acknowledgements
This study was supported in part by a Basic Science
research grant SC2002 ⁄ 0502 from Enterprise Ireland
and this assistance is gratefully acknowledged. We also
wish to thank Roche Diagnostics and the Irish Ameri-
can Partnership for studentship support in the initial
stages of the project and the EU for making it possible

sophila alcohol dehydrogenase increases cofactor speci-
ficity for NADP
+
. FEBS Lett 356, 81–85.
8 Scrutton NS, Berry A & Perham RN (1990) Redesign
of the coenzyme specificity of a dehydrogenase by pro-
tein engineering. Nature 343, 38–43.
9 Baker PJ, Britton KL, Engel PC, Farrants GW, Lilley
KS, Rice DW & Stillman TJ (1992) Subunit assembly
and active site location in the structure of glutamate
dehydrogenase. Proteins Struct Funct Gen 12, 75–86.
10 Smith EL, Austen BM, Blumenthal KM & Nyc JF
(1975) Glutamate dehydrogenases. In The Enzymes,
Vol. 11 (Boyer PD, ed), pp. 293–367. Academic Press,
New York, NY.
11 Hudson RC & Daniel RM (1993) L-glutamate dehydro-
genases: distribution, properties and mechanism. Comp
Biochem Physiol B 106, 767–792.
12 Teller JK, Smith RJ, McPherson MJ, Engel PC &
Guest JR (1992) The glutamate dehydrogenase gene of
Clostridium symbiosum. Cloning by polymerase chain
reaction, sequence analysis and over-expression in
Escherichia coli. Eur J Biochem 206, 151–159.
13 Smith TJ, Peterson PE, Schmidt T, Fang J & Stanley
CA (2001) Structures of bovine glutamate dehydroge-
nase complexes elucidate the mechanism of purine regu-
lation. J Mol Biol 320, 707–720.
14 Buckel W & Barker HA (1974) Two pathways of gluta-
mate fermentation by anaerobic bacteria. J Bacteriol
117, 1248–1260.

of non-fusion proteins. Nucleic Acids Res 14, 3603.
23 Kaplan NO, Ciotti MM & Stolzenbach FE (1957) Stud-
ies on the interaction of diphosphopyridine nucleotide
analogs with dehydrogenases. Arch Biochem Biophys 69,
441–457.
24 Wilkinson GN (1961) Statistical estimations in enzyme
kinetics. Biochem J 80, 324–332.
Changing glutamate dehydrogenase coenzyme specificity J. B. Carrigan and P. C. Engel
5174 FEBS Journal 274 (2007) 5167–5174 ª 2007 The Authors Journal compilation ª 2007 FEBS