Báo cáo khoa học: Correlation between conformational stability of the ternary enzyme–substrate complex and domain closure of 3-phosphoglycerate kinase potx

Correlation between conformational stability of the
ternary enzyme–substrate complex and domain closure
of 3-phosphoglycerate kinase
Andrea Varga
1
, Bea
´
ta Flachner
1
,E
´
va Gra
´
czer
1
, Szabolcs Osva
´
th
2
, Andrea N. Szila
´
gyi
1
and Ma
´
ria Vas
1
1 Institute of Enzymology, Biological Research Center, Hungarian Academy of Sciences, Budapest, Hungary
2 Department of Biophysics and Radiation Biology, Semmelweis University, Budapest, Hungary
The experimental and theoretical studies that led to
our present understanding of protein structural chan-

ges and their role in enzyme function were mostly
carried out on small single-domain proteins [1]. Most
enzymes, however, are built of several domains. The
mechanism of transmission of molecular interactions
over large distances (e.g. from one domain to the
other) within the molecule, the role of substrates in
Keywords
domain closure; phosphoglycerate kinase;
molecular graphics; substrate effect;
thermal unfolding
Correspondence
M. Vas, Institute of Enzymology, BRC,
Hungarian Academy of Sciences, H-1518
Budapest, PO Box 7, Hungary
Fax: +36 1466 5465
Tel: +36 1279 3152
E-mail:
(Received 18 December 2004, revised 15
February 2005, accepted 17 February 2005)
doi:10.1111/j.1742-4658.2005.04618.x
3-Phosphoglycerate kinase (PGK) is a typical two-domain hinge-bending
enzyme with a well-structured interdomain region. The mechanism of
domain–domain interaction and its regulation by substrate binding is not
yet fully understood. Here the existence of strong cooperativity between
the two domains was demonstrated by following heat transitions of pig
muscle and yeast PGKs using differential scanning microcalorimetry and
fluorimetry. Two mutants of yeast PGK containing a single tryptophan
fluorophore either in the N- or in the C-terminal domain were also studied.
The coincidence of the calorimetric and fluorimetric heat transitions in all
cases indicated simultaneous, highly cooperative unfolding of the two

domains. This cooperativity is preserved in the presence of substrates:
3-phosphoglycerate bound to the N domain or the nucleotide (MgADP,
MgATP) bound to the C domain increased the structural stability of the
whole molecule. A structural explanation of domain–domain interaction is
suggested by analysis of the atomic contacts in 12 different PGK crystal
structures. Well-defined backbone and side-chain H bonds, and hydropho-
bic and electrostatic interactions between side chains of conserved residues
are proposed to be responsible for domain–domain communication. Upon
binding of each substrate newly formed molecular contacts are identified
that firstly explain the order of the increased heat stability in the various
binary complexes, and secondly describe the possible route of transmission
of the substrate-induced conformational effects from one domain to the
other. The largest stability is characteristic of the native ternary complex
and is abolished in the case of a chemically modified inactive form of
PGK, the domain closure of which was previously shown to be prevented
[Sinev MA, Razgulyaev OI, Vas M, Timchenko AA & Ptitsyn OB (1989)
Eur J Biochem 180, 61–66]. Thus, conformational stability correlates with
domain closure that requires simultaneous binding of both substrates.
Abbreviations
1,3-BPG, 1,3-bisphosphoglycerate; 3-PG, 3-phospho-
D-glycerate; CM-PGK, carboxamidomethylated PGK;DSC, differential scanning
microcalorimetry; GAPDH,
D-glyceraldehyde-3-phosphate dehydrogenase; PGK, 3-phospho-D-glycerate kinase or ATP:3-phospho-D-glycerate
1-phosphotransferase; W122, yeast PGK with mutations of Y122W, W308F and W333F; W333, yeast PGK with mutation of W308F.
FEBS Journal 272 (2005) 1867–1885 ª 2005 FEBS 1867
this process, and the fulfilment of enzyme activity
through these structural changes remain to be
elucidated.
3-Phosphoglycerate kinase (PGK; EC 2.7.2.3) is a
typical two-domain hinge-bending enzyme [2–5] with a

conserved primary structure [6] and tertiary fold [2–
5,7–9], including a well-structured interdomain region.
PGK is therefore a suitable model with which to study
the mechanism of domain–domain interplay and its
role in both protein stability and function. In order to
understand how the two substrates affect domain
interplay and thereby induce the hinge-bending type of
domain movement, the effect of the substrates (both
separately and together) on PGK conformation needs
to be clarified.
Substrate-induced stabilization of PGK structure
against chemical modification [10–15], proteolytic de-
gradation [16,17] and unfolding [12,18–20] has been
observed. Substrate-induced conformational changes
have been detected by various techniques, inclu-
ding NMR [21–23], fluorescence [15,24–29] and ESR-
spectroscopy [30], analytical ultracentrifugation [31],
small angle X-ray scattering [32–34] and X-ray crystal-
lography [3,4,8]. There are, however, conflicting results
concerning the requirement of either only one or both
substrates in causing domain movement and the ques-
tion of whether binding of a single substrate to one of
the two domains can also stabilize the structure of the
other domain [32–34]. Often the results of studies with
the solubilized PGK do not support the suggestion
from the X-ray crystallographic work about the
requirement of both substrates (i.e. formation of the
ternary enzyme–substrate complex) for domain closure.
The contradiction may be attributed to prevention of
occurrence of the large-scale movement of domain clo-

sure by the lattice forces operating in certain PGK
crystals [35].
The best approach by which to obtain direct infor-
mation about the extent of domain–domain coopera-
tivity, is to carry out unfolding–refolding experiments,
as noninteracting structural domains generally corres-
pond to separate folding units. Both unfolding [36–50]
and refolding [36–38,46–62] experiments using denatu-
rants have been carried out extensively with PGK, but
mainly in the absence of substrates, with the intact
enzyme [36,37,39,41,43,44,46,48], its engineered mutants
[39–41,46,55,62] and its various molecular fragments
[40,50–55,57–63]. These studies show that the two
domains exhibit slightly different stabilities and unfold-
ing ⁄ refolding of the two domains probably occurs in a
sequential order within the PGK molecule. No com-
prehensive picture, however, has yet emerged about
the relative stability of the two domains and about the
extent of domain–domain interactions, especially not
in the enzyme-substrate complexes.
Heat- or cold-induced unfolding of yeast [19,20,64–
73], thermophile [64,65,74,75] and cold-active [76]
PGKs in the absence [64–75] or presence [19,20,76] of
substrates has been monitored by differential scanning
calorimetry (DSC), a widely used approach to deter-
mine the number of folding units within a protein
molecule. Disruption of native PGK structure upon
cooling occurs in two distinct stages, corresponding to
independent and reversible unfolding of the individual
domains [66,70–72]. Recently it has been suggested

that the uncoupled unfolding of the two domains is a
result of the presence of relatively high concentration
of guanidine hydrochloride used in cold denaturation
experiments [62]. On the other hand, thermal unfolding
of PGK invariably proceeds in an apparently single
DSC transition both at low guanidine hydrochloride
concentration [69] and in diluted buffer [19,20,67].
Strong interdomain stabilization has been claimed in
both cases, but the slightly asymmetric thermal unfold-
ing profile may be accounted for by assuming partially
separated unfolding of the N- and C-terminal domains.
For the protective effect of substrates against thermal
unfolding of PGK the experimental data are scarce
[19,20,76], no systematic comparison has been made
in the various binary and ternary enzyme–substrate
complexes.
Separated DSC transitions attributable to the indi-
vidual PGK domains have been observed in the case
of engineered mutants of yeast PGK that contain modi-
fications in the hinge region between the two domains
[20,68] and in the case of cold-active PGK [76], pos-
sibly due to weaker interdomain interactions in these
cases. It is notable that the substrate 3-phospho-d-
glycerate (3-PG) in one case [68], while 3-PG and
MgADP together in the other case [76] caused merging
of the two transitions into a single one. This indicates
increased domain cooperativity upon substrate binding
in the mutants. Another thermal unfolding study of a
substrate-free PGK from the thermophilic bacterium
Thermotoga maritima has led to the proposal of a

four-state model with three well-defined unfolding
transitions: disruptions of domain–domain interactions
and subsequent sequential unfolding of the two
domains [75].
Domain coupling and dependence on substrate bind-
ing are important not only in the stabilizing mechan-
ism of PGK, but also in interdomain communication
during the catalytic cycle. To characterize the extent of
domain coupling and its regulation by each substrate
we have devised thermal unfolding experiments with
the mammalian pig muscle PGK, studied in our
Substrate-assisted domain–domain cooperativity A. Varga et al.
1868 FEBS Journal 272 (2005) 1867–1885 ª 2005 FEBS
laboratory. For comparison, yeast PGK was also
investigated. The purpose was twofold: to resolve, as
much as possible, the overlapping thermal transitions
of the two domains; and to determine the effects of
substrates (in their binary and ternary complexes with
PGK) on the thermal transitions. To achieve this, heat
transitions of wild-type pig muscle PGK as well as that
of the wild-type and two single Trp mutants of yeast
PGK were monitored by applying two independent
methods, microcalorimetry and fluorimetry. The try-
ptophans of the two mutants are located either in the
N- or in the C-terminal domain [39,41], allowing
selective fluorimetric detection of the conformational
changes within the domains, while DSC calorimetry
characterizes the two domains together. Furthermore,
the effects of substrates on thermal unfolding of PGK
are compared in various binary and ternary complexes

and we investigated whether these effects correlate with
the existing molecular interactions, known from the
X-ray structures.
Results and Discussion
Interdomain interactions of PGK
Coincidence of calorimetric and fluorimetric heat
transitions of wild-type PGKs reflects domain
co-operativity
To test the extent of domain–domain coupling during
thermal unfolding of PGK, we performed both DSC
and fluorimetric heat transition experiments with the
wild-type pig muscle and yeast PGKs. In both
enzymes, the Trp residues, mainly responsible for the
protein fluorescence, are located within the C-terminal
domain: four Trp-s in pig [5], and two Trp-s in the yeast
PGKs [7]. Thus, if there is any uncoupling between the
domains, a noncoincidence of the calorimetric and
fluorimetric transition temperatures is expected, even if
their thermal unfolding is not well separated.
Typical heat capacity plots of DSC experiments and
fluorimetric thermal transition curves obtained for the
thermal unfolding of the substrate-free pig muscle
PGK and its complexes with various substrates are
shown in Fig. 1B and C. Part of the curves in Fig. 3
illustrates similar experiments with wild-type yeast
PGK. The T
m
values and the experimental calorimetric
heats of unfolding (Q
t

) are given in Table 1, indicating
pronounced protection by the substrates, details of
which will be discussed later.
In all cases the DSC transition curves are character-
ized by single, slightly asymmetric transitions. No
residual structure could be detected after the heat
transition by far UV CD spectroscopy (data not
shown). Heat denaturation of pig muscle PGK is an
irreversible process, similar to that of the yeast enzyme
[19,67]. No repeated heat transition was observed on
subsequent re-scanning of the sample. Furthermore,
the observed T
m
values are found to be strongly scan-
rate-dependent, thus, they are at least partially under
kinetic control, similar to those of yeast PGK [67].
These findings provide evidence of a nonequilibrium
unfolding mechanism. The irreversible nature of the
50
40
30
20
10
0
40
30
20
10
0
1.2

1.0
0.8
0.6
0.4
0.2
0.0
40 45 50 55 60 65
Temperature (°C)
F
N
∆C
P
(kcal/mole/°C)

∆C
P
(kcal/mole/°C)

A
B
C
Fig. 1. Effect of substrates on the temperature-dependent unfold-
ing of pig muscle PGK. DSC (A, B) and fluorimetric (C) heat dena-
turation curves were determined at the scanning rate of
1.0 KÆmin
)1
with carboxamidomethylated (A) and unmodified (B, C)
pig muscle PGK in the absence of substrates (d), in the presence
of 10 mM MgATP (n), 10 mM MgADP (m), 10 mM 3-PG (h)and
10 mM MgADP plus 10 mM 3-PG (r). In (A) and (B) the values of

excess heat capacity (DC
P
) is plotted against temperature.
A. Varga et al. Substrate-assisted domain–domain cooperativity
FEBS Journal 272 (2005) 1867–1885 ª 2005 FEBS 1869
heat transition is also supported by the simultaneous
increase of light scattering during the heat transition
of PGK (data not shown), which indicates occurrence
of accompanying aggregation. The mechanism is poss-
ibly a complex one since deviation from the simple
two-state irreversible native fi unfolded denaturation
model has been claimed from previous DSC experi-
ments with yeast PGK [67]. We have also analysed the
present DSC data according to the kinetic model for
irreversible denaturation as described by Sanchez-Ruiz
et al. [77]. The results in Fig. 2A show that, although
the data can be approximated by straight lines, they
indeed, deviate from the simple two-state irreversible
denaturation model, as no common straight line is
formed when different scan rates are applied [78].
Our main purpose, however, was not the clarifica-
tion of the overall mechanism of thermal unfolding.
We have restricted ourselves to the question of whe-
ther unfolding of the two domains is (even slightly)
separated. Therefore, thermal transitions were also
followed by measuring tryptophan fluorescence inten-
sity changes (Fig. 1C). It should be noted that these
transitions are not distorted by the above-mentioned
aggregation, since the extent of aggregation is propor-
tional to the changes in protein fluorescence during the

whole transition. Under identical scan rate and condi-
tions, within the experimental error, the same T
m
values were observed in fluorescence (Fig. 1C) as in the
DSC-experiments (Fig. 1B) and the data are summar-
ized in Table 1. The coincidence of the T
m
values deter-
mined by fluorimetry and calorimetry indicates that
disruption of the C-domain structure and of the whole
molecule cannot be separated, i.e. the two domains are
possibly disrupted in a highly co-operative way.
Different heat stabilities of two single Trp mutants
of yeast PGK are not related to domain uncoupling
In the above experiments thermal unfolding of either
the whole PGK molecule or its C-terminal domain
(within the intact molecule) was monitored. In order
to detect unfolding of both N and C domains within
the molecule more directly, we performed comparative
DSC and fluorimetric heat transition experiments on
two single Trp mutants of yeast PGK. The Trp residue
was either in the N- (W122) or the C-terminal (W333)
domain, as described by Mas et al. [39,41]. Residue
numbers 122 and 333 correspond to 123 and 335 in
pig muscle PGK, respectively. In agreement with previ-
ous data [39,41], we found that these mutants are fully
active, thus, the mutations do not perturb the structure
significantly. Stabilities of the mutants are also only
slightly decreased with respect to the wild-type enzyme
in guanidine hydrochloride-induced denaturation [41,

62]. Previously a sequential domain-unfolding model
has been suggested for the mutants. According to this
model the C domain unfolds first, while the N domain
remains relatively compact, but looses most of its ter-
tiary structure. Complete unfolding of the N domain
occurs only during the second transition [41]. The
reverse order of stabilities, however, has been reported
for the isolated N- and C-terminal domains of the
mutants [62]. Therefore, if the two PGK domains
unfold sequentially (in either order) not only in the
denaturant-induced, but also in heat denaturation
transitions, a different type of noncoincidence of the
calorimetric and fluorimetric transition temperatures is
expected with the W122 and the W333 mutants,
respectively.
Table 1. Mid-point temperatures (T
m
) and calorimetric heats (Q
t
) of thermal transitions of pig muscle and yeast PGKs. T
m
and Q
t
values are
given in °C and kcalÆmol
)1
, respectively. T
m,cal
and T
m,fluor

were determined by DSC and fluorimetric experiments, respectively, as shown in
Figs 1 and 3. The experimental errors of T
m
and of Q
t
were ± 0.2–0.3 °C and ± 5–10 kcalÆmol
)1
, respectively.
Ligand
Pig muscle PGK Yeast PGK
Unmodified CM- Wild-type W122 W333
T
m,cal
T
m,fluor
Q
t
T
m,cal
Q
t
T
m,cal
T
m,fluor
Q
t
T
m,cal
T

m,fluor
Q
t
T
m,cal
T
m,fluor
Q
t
No 53.0 53.1 114 46.5 85 56.4 56.2 110 47.5 48.5 49 52.6 52.9 103
MgATP 54.8 55.3 125 47.6 91 59.2 59.8 127 49.4 51.3 60 56.0 55.7 126
55.6
a
–––––––––––––
MgADP 56.9 57.0 139 48.7 98 60.7 61.4 140 51.5 52.3 73 57.0 57.5 130
57.6
a
–––––––––––––
3-PG 58.2 57.5 147 52.4 110 60.1 60.0 150 51.4 51.7 79 56.6 57.2 131
3-PG + MgADP 59.1 58.6 175 51.1 105 62.8 63.4 157 53.7 54.0 89 58.9 59.6 150
59.8
b
–––––––––––––
a
Published values, taking into account the decreasing effect of Mg
2+
on PGK stability [83].
b
Extrapolated to zero concentration of free
Mg

2+
using the method described earlier [83].
Substrate-assisted domain–domain cooperativity A. Varga et al.
1870 FEBS Journal 272 (2005) 1867–1885 ª 2005 FEBS
The most important feature of the results is the
good correlation of the experimental T
m
values
(Table 1), determined either calorimetrically (Fig. 3A)
or fluorimetrically (Fig. 3B). These results strongly
argue in favour of a highly co-operative thermal
unfolding of the two domains in case of both enzyme
forms, although the stabilities of the whole molecules
differ from each other and from that of the wild-type
yeast PGK.
Quantitative analysis of the heat transition data, on
one hand, gave straight lines for W333 and wild-type
PGKs (Fig. 2B), in agreement with the kinetic model
for irreversible denaturation [77]. Non-coincidence of
the calorimetric and fluorescence data, however,
indicates deviation from the one-step irreversible
native fi unfolded denaturation model (similar to the
findings in Fig. 2A). This deviation may influence
slightly differently the calorimetric and fluorimetric
detection of unfolding. For the substrate-free W122
mutant, on the other hand, deviation from the two-
state irreversible model is very pronounced. This is
indicated by both the biphasic nature of the curve in
Fig. 3B and the well visible deviation from straight
lines, especially in case of the more sensitive fluorimet-

ric method (Fig. 2B). This behaviour of the W122
mutant is consistent with the previously suggested two-
step unfolding of the N domain by Mas’ group [40,41].
As this study shows that two-step behaviour is
observed for both the calorimetric and fluorimetrically
detected heat transitions of the intact molecule, it is
conceivable that melting of the two domains, even in
this case occurs in a highly co-operative way.
The co-operative unfolding mechanism, shown from
the experiments with either wild-type or mutant PGKs,
differs largely from the sequential mechanism derived
previously for refolding of the two domains of pig
muscle PGK [60]. This is probably due to the fact
that ) in contrast with refolding ) unfolding starts
from the native structure with folded domains and
established interdomain interactions, resulting in a
stronger coupling between domains.
Structural basis of PGK domain cooperativity:
conserved features of the interdomain region
In order to rationalize the structural basis of the highly
cooperative domain–domain interactions, indicated by
previous and present calorimetric data, we were look-
ing for the similarities in 12 available crystal structures
of various PGKs. Three different types of molecular
contacts were collected and visually investigated: (a)
backbone peptide H bonds; (b) electrostatic and
H-bonding contacts; (c) hydrophobic interactions
between side chains of the conserved residues or
between atoms of backbone peptides and of the con-
served side chains. From these molecular contacts only

those that exist in all PGK structures were selected,
independently of the source, the conformational state
(open or closed) and of ligation with substrates.
Among the backbone H bonds (Fig. 4A) there are
special ones (listed in Table 3) which are in crucial
positions, directly linking the nearby secondary struc-
tural elements to the previously described C- and
N-terminal hinges of the interdomain helix 7 [3] as well
as to bL, where the main hinge is possibly located [5]
(Fig. 4B and C).
-4
-5
-6
-7
-8
-9
-10
3.04
3.00
1.0
0.5
0.0
-0.5
-1.0
-1.5
3.05 3.10 3.15 3.20
10
3
/T (K
-1

)
In(In[N]
0
/[N]) or In(In(Q
t
/(Q
t
-Q)))
In(k/T)
3.06 3.08 3.10 3.12
A
B
Fig. 2. Linear transformation of the thermal unfolding data. Plots
(A) and (B) were prepared by using Eqns (7) and (8), respectively.
In (A) the data of the DSC transition curves with the substrate free
pig muscle PGK, obtained at scanning rates (v) of 1.5 (n), 0.7 (h),
0.4 (s) and 0.1 (Ñ)KÆmin
)1
, were analysed. (B) Data of calorimetric
(filled symbols) and fluorimetric (unfilled symbols) measurements,
using the scanning rate of 1.0 KÆmin
)1
, were compared for wild-
type pig muscle (d,s), yeast (j,h) and W122 (r,e), W333 (m,n)
mutant yeast PGKs in the absence of substrates. The original data
are shown in Figs. 1B, 1C, 3A and 3B. The activation parameters
obtained from plot (A) agreed within the experimental error with
the values given in Table 2.
A. Varga et al. Substrate-assisted domain–domain cooperativity
FEBS Journal 272 (2005) 1867–1885 ª 2005 FEBS 1871

The H bonds at the N-hinge (Fig. 4C) create a con-
nection between the C and N terminals of the polypep-
tide chain (a special structural feature of the molecule),
as well as between helices 5 and 7. Here we emphasize
the conservative nature of the H bonds stabilizing this
region and their importance in determining the posi-
tion of the whole N domain relative to the interdo-
main helix 7. A similar role can be attributed to the H
bonds at the C-hinge (Fig. 4B). As we show below, in
addition to these permanent bonds, there are further,
changeable H bonds. The number and exact location
of these bonds vary upon ligation with the substrates
and with the conformational states of the protein
molecule. The changeable H bonds may contribute to
stabilization of the various conformational states,
while the permanent ones may allow a rigid-body-like
movement of the domains relative to helix 7.
There are no special H bonds within the inter-
domain region, but the entire region is built up of
conserved residues. Their hydrophobic (Fig. 4D),
electrostatic and H bonding (Fig. 4E) interactions are
listed in Table 3. These contacts of the conserved resi-
dues exist in all PGK structures, independent of their
conformational states (open or closed) or the complex
formation with various substrates or ligands. An exten-
ded hydrophobic cluster dominates in the interdomain
interactions. The known cold sensitivity of these forces
may correlate with the finding that unfolding of the
two domains is not coupled during cold denaturation
[66,73]. This hydrophobic cluster together with the

ionic interactions and H bonds constitute a well organ-
ized interdomain region which may have great import-
ance both in mediating conformational effects between
the domains and in unifying the two domains into a
single cooperative melting unit at elevated tempera-
tures.
Stabilization of PGK conformation in the binary
substrate complexes
Protection by the individual substrates against thermal
unfolding
Both the T
m
-values and the experimental calorimetric
heats (Q
t
) required for unfolding (Table 1) are
increased significantly by the substrates, indicating
their distinct protective effects on PGK conformation
Temperature (°C) Temperature (°C)
F
N
∆C
P
(kcal/mole/°C)

403530
30
25
20
15

10
5
0
1.0
0.8
0.6
0.4
0.2
0.0
45 50 55 60
4035 45 50 55 60 65
A
BD
C
Fig. 3. Heat transitions of wild-type and mutant yeast PGKs. DSC (A, C) heat denaturation curves of wild-type (j), W122 (r) and W333 (m)
yeast PGKs and fluorimetric (B, D) heat denaturation curves of wild-type (h), W122 (e) and W333 (n) yeast PGKs were determined at the
scanning rate of 1.0 KÆmin
)1
, in the absence of substrates (A, B) and in the presence of 10 mM 3-PG (C, D). In (A) and (C) the values of
excess heat capacity (DC
P
) is plotted against the temperature.
Substrate-assisted domain–domain cooperativity A. Varga et al.
1872 FEBS Journal 272 (2005) 1867–1885 ª 2005 FEBS
(Fig. 1). Due to the irreversibility of the folding process,
characterization of the substrate-caused effects accord-
ing to the equilibrium thermodynamics is not possible.
The deviation form the two state irreversible model is,
however, apparently not too large in most cases
(Fig. 2). Thus, the reversibly unfolded intermediate

Fig. 4. Non-covalent bonds of PGK responsible for interdomain interactions. The ribbon diagram of the open conformation of the substrate-
free pig muscle PGK [80] (A) and its details at the C-hinge (B), at the N-hinge (C) as well as in the interdomain region including the main
hinge at bL (D and E) are shown. Various important secondary structure elements are labelled and coloured differently. In figures A, B and
C, the backbone of the polypeptide chain is also illustrated (stick model) together with the stabilizing H-bonds (dashed lines). The conserved
side-chains (stick models) in the interdomain region are seen with their hydrophobic (D), electrostatic and H-bonding (E) interactions (dashed
lines). The interaction distances are listed in Table 3.
A. Varga et al. Substrate-assisted domain–domain cooperativity
FEBS Journal 272 (2005) 1867–1885 ª 2005 FEBS 1873
state(s) may not accumulate in detectable amounts, i.e.
their formation is possibly much slower than their
decay into an irreversibly unfolded state. On this basis,
except for the ligand-free W122 mutant of yeast PGK,
we could estimate the kinetic activation parameters of
the process from the type of plot shown in Fig. 2A.
These parameters are summarized in Table 2. The sub-
strates increase both the activation enthalpy (DH
à
) and
the activation entropy (DS
à
) of unfolding in a way that
finally leads to an increase of the activation free
enthalpy (DG
à
), which quantitatively measures the sta-
bilization effect. Of substrates studied 3-PG had the
strongest, MgADP an intermediate and MgATP the
weakest stabilizing effect. The observed order of stabil-
ity of the various PGK–substrate complexes is inter-
preted below on structural basis.

It is also evident from these results that a similar
cooperative mechanism operates in the binary com-
plexes with either substrate, i.e. their stabilizing effect
is not restricted to the N or C domain, respectively, to
which they bind. Thus, each of the substrates also sta-
bilizes the domain to which they do not bind, provi-
ding further evidence in favour of operation of strong
domain–domain interactions.
Molecular explanation of the increased conformational
stability by 3-PG
The largest protection among the investigated sub-
strates against thermal unfolding of PGK was shown
by 3-PG, which is in agreement with spectroscopic
studies [29].
To describe the effect of substrates in structural
terms, we have searched for new atomic contacts
within the protein formed only upon substrate binding.
The contacts between the bound 3-PG and PGK,
known from crystallographic studies of this binary
complex of pig muscle PGK [8] (Fig. 5A) as well as
from other 3-PG containing PGK structures [3–5,79–
81] suggest a possible way of stabilization by 3-PG.
Namely, all side chains interacting with 3-PG belong
to separate structural elements of the N domain,
namely bA, bB, bD and bE as well as helices 1 and 5.
Thus, these structural elements are strongly fixed
together by 3-PG and this may result in an increased
stability of the whole N domain.
Calorimetric experiments, however, indicate that
3-PG and other substrates stabilize the whole mole-

cule, not only the domain to which they bind. This
effect is most probably promoted by the existing inter-
actions between the two domains. The transmission of
3-PG induced effects from the N domain to the C
domain may be visualized by observing the newly
formed interactions within the protein molecule (col-
oured violet in Fig. 5A, Table 3), characteristic of all
3-PG-bound PGK structures. The Arg38 side chain
(helix 1) makes new H bonding with Thr393 (peptide
O atom) in bL, and by the aid of the permanent elec-
trostatic interaction with the carboxylate of Asp23
(bA) a new connection is formed between the two
domains. Numbering of residues throughout the text
refers to pig muscle PGK sequences, unless stated
otherwise e.g. Bacillus stearothermophilus (Bs) or Try-
panosoma brucei (Tb). This connection is characteristic
of all 3-PG-bound structures [3,4,81,82], whereas bind-
ing of MgADP or MgATP to the substrate-free
enzyme does not induce formation of this bond
(Table 3).
Based on structural comparison we present here a
possible mechanism of the conformational changes
caused by 3-PG that lead to the formation of the
Arg38–Thr393 interaction. Upon 3-PG binding the
strong interaction between its phosphate and the side
chain of Arg170 on helix 5 shifts the whole helix 5,
Table 2. The activation parameters of thermal transitions of pig muscle and yeast PGKs. The values of DH
à
(kcalÆmol
)1

)andDS
à
(calÆmol
)1
)
were derived from the type of plot shown in Fig. 2A, prepared from DSC measurements. DH
à
was assumed to be independent of the tem-
perature within the range of the measurements. The calculated DG
à
(kcalÆmol
)1
) values are referred to 25 °C. The errors of DH
à
and DS
à
are
± 10–15%.
Ligand
Pig muscle PGK Yeast PGK
Unmodified CM- Wild type W122 W333
DH
à
DS
à
DG
à
DH
à
DS

à
DG
à
DH
à
DS
à
DG
à
DH
à
DS
à
DG
à
DH
à
DS
à
DG
à
No 129 329 31.0 128 333 28.9 124 310 32.0 – – – 165 438 33.9
MgATP 160 418 35.1 137 359 29.9 131 326 33.5 120 303 29.1 197 530 38.5
MgADP 204 551 39.6 127 328 29.4 149 380 35.8 138 357 31.3 190 509 38.2
3-PG 174 457 37.3 122 307 30.6 199 530 40.8 161 428 33.4 224 612 41.6
3-PG + MgADP 241 658 44.9 132 339 30.8 247 671 47.5 174 465 35.3 242 663 44.7
Substrate-assisted domain–domain cooperativity A. Varga et al.
1874 FEBS Journal 272 (2005) 1867–1885 ª 2005 FEBS
thereby the ring of Phe165 (see Fig. 4D) is also dis-
placed parallel to its former position by at least about

2A
˚
. This effect is further enhanced through the inter-
action with Glu192 (helix 7) and causes an additional
shift of about 3.5 A
˚
in the position of the imidazole ring
of the interacting His390 (Fig. 4E). Since His390 is
located in bL, the conformation of this b-strand may
also become significantly altered. This conformational
change would be directly related to the domain move-
ment, since there are strong arguments supporting that
the main molecular hinge of PGK is located in bL [5].
This may be the explanation for the small extent of
domain rotation observed in the 3-PG binary complex
[8]. The new conformation of bL is stabilized by a new
H bond between the conserved Ser392(OG) and
Gly394(N) (Fig. 5A), characteristic of all PGK struc-
tures which bind 3-PG. These changes also lead to a
roughly 2 A
˚
shift of the backbone atoms of Thr393
towards the guanidium group of Arg38 and the two
may reach each other within H-bonding distance. The
conformational change caused by 3-PG binding in bL
can be further transmitted to the C domain through the
H-bonding system between bL and bK shown in
Fig. 4B. During this process formation of new H
bonds between Thr375(OG1) (from helix 13 that is
sequentially situated between bK and bL) and

Gly337(N), as well as between the nearby Val339(N)
and Thr351(OG1) (belonging to helix 12) may streng-
then the interactions within the C domain (Fig. 5A). It
is notable, that Gly337 and Val339 are in a loop
between bJ and helix 12, which is just below the binding
site of the nucleotide substrate. These latter H bonds
also exist in the binary complex with the nucleotide sub-
strates, but they are absent in the substrate-free PGK.
Through the contacts described above, the conform-
ational changes induced by substrate binding can be
transmitted from the 3-PG-site of the N domain to the
nucleotide binding pocket of the C domain leading to
stabilization the whole enzyme molecule.
Fig. 5. Details of the interdomain region in the binary complexes
with substrates. (A) and (C) show 3-PG and MgATP binding,
respectively, to pig muscle PGK [8,83], while MgADP binding to
B. stearothermophilus PGK [9] is shown in (B). In each case
sequence numbering refers to the corresponding species. The
important secondary structural elements are highlighted as ribbons
with the same colour as shown in Fig. 4. Blue ball and stick models
represent the bound substrates. Only the side-chain or backbone
atoms (stick models) interacting with the substrates and the ones
forming new interactions in the protein molecule, characteristic of
the substrate-bound structures are shown and coloured violet. The
interacting atoms are connected with dashed lines, while arrows
connect the equivalent noninteracting atoms in C. In the latter
case, the distances are also indicated in angstroms. One perma-
nent peptide H bond (370:O)392:N in A or Bs348:O–Bs370:N in B)
that makes connection between bK (green) and bL (red), character-
istic of all PGK structures, is also indicated (dashed lines). The

protein contacts together with the distances between the corres-
ponding atoms in the substrate-free and the closed ternary com-
plex structures are listed in Table 3.
A. Varga et al. Substrate-assisted domain–domain cooperativity
FEBS Journal 272 (2005) 1867–1885 ª 2005 FEBS 1875
Table 3. Atomic contacts responsible for domain cooperation. Atomic distances (A
˚
-s) were measured in the interdomain region and its sur-
roundings. From the total of 12 crystal structures investigated, data of those most characteristic were selected. The contacts that vary upon
substrate binding or upon domain closure are indicated in bold. The contact list for the closed structure of T. maritima PGK (data not shown)
is similar to that of T. brucei PGK, with few exceptions.
Interacting
structural
elements
Pig muscle PGK (open) B. stearotherm. PGK (open) T. brucei PGK (closed)
Atom 1 Atom 2
Substrate
free
3-PG
binary
MgATP
binary Atom 1 Atom 2
MgADP
binary Atom 1 Atom 2 Ternary
c
Backbone H bonds
N- and C-terminals N4:O V416:N 3.00 2.98 2.46 N2:O K394:N 2.74 Q5:O K419:N 2.84
L6:N S414:O 2.74 2.78 2.66 K4:N Q392:O 2.56 K7:N D417:O 2.79
N-terminus and bF L6:O G185:N 2.98 2.78 2.85 K4:O A167:N 3.00 K7:O A189:N 2.84
L8:N G185:O 2.66 2.92 2.94 I6:N A167:O 3.22 I9:N A189:O 3.22

Helix 5 and C terminus A168:O G409:N 2.83 2.81 2.69 A149:O G387:N 2.76 A170:O G412:N 2.72
bF and C terminus K183:O S414:N 2.68 2.98 2.83 A165:O E392:N 2.96 G187:O D417:N 2.90
Within helix 7
(C-terminal part)
A199:O
a
E201:N 4.42 3.60 4.37 A181:O S183:N 3.89 V203:O G205:N 3.22
A199:O
a
P203:N 5.54 4.19 5.67 A181:O P185:N 4.07 V203:O P207:N 3.32
S202:O
a
E204:N 4.03 3.89 4.08 N184:O D186:N 3.74 N206:O P208:N 3.41
P203:O
a
R205:N 3.80 3.75 4.13 P185:O R187:N 3.77 P207:O R209:N 3.25
Helix 7 (C-term.) and bJ E204:O
a
N332:N 4.44 4.14 4.50 D186:O L312:N 4.16 P208:O K334:N 3.27
bG and bH F207:O N231:N 2.83 2.71 2.98 F189:O D213:N 2.79 L211:O D235:N 2.73
A209:O I234:N 2.84 2.91 2.87 A191:O I216:N 2.86 A213:O L238:N 2.82
bG and bJ P206:O K331:N 2.71 2.85 2.88 P188:O K311:N 2.82 P210:O K333:N 2.83
P206:O Q332:N 3.25 3.35 3.14 P188:O L312:N 3.37 P210:O S334:N 3.04
L208:N Q332:O 2.89 2.89 2.83 T190:N L312:O 2.99 V212:N S334:O 2.97
I210:O N336:N 2.95 2.92 2.94 I192:O N316:N 2.96 I214:O N338:N 2.85
bKandbL I370:O S392:N 2.97 2.90 2.97 I348:O S370:N 2.90 I373:O S395:N 2.70
G372:N
a
S392:O 4.62 4.16 4.15 G350:N S370:O 5.36 G375:N S395:O 3.46
G372:N

a
G394:O 4.47 4.90 3.98 G350:N G372:O 7.20 G375:N G397:O 3.24
Within helix 13 C379:N
b
T375:O 3.82 3.20 3.41 V357:N S353:O 3.18 A382:N S378:O 2.96
A380:N
b
A376:O 3.88 3.45 3.73 E358:N A354:O 3.05 Q383:N A379:O 2.96
Hydrophobic interactions
Helix 5 and helix 7 F165:CG E192:CG 3.64 3.58 3.69 F146:CG E174:CG 3.86 F167:CG E196:CG 3.53
F165:CD2 E192:CB 3.68 3.54 3.74 F146:CD2 E174:CB 3.85 F167:CD2 E196:CB 3.38
F165:CE1 E192:CD 3.72 3.85 4.08 F146:CE1 E174:CD 4.47 F167:CE2 E196:CD 3.72
F165:CZ F196:CE1 3.73 4.19 4.15 F146:CZ L178:CD1 3.72 F167:CZ F200:CE1 3.36
Helix 5 and helix 14 F165:CD1 G394:CA 3.98 3.64 4.06 F146:CD1 G372:CA 3.65 F167:CD2 G397:CA 4.01
F165:CZ A397:CB 4.06 4.35 4.36 F146:CZ A375:CB 4.69 F167:CZ A400:CB 4.41
F165:CZ S398:CA 3.91 4.50 4.07 F146:CZ S376:CA 4.16 F167:CZ S401:CA 4.40
F165:CZ L401:CD1 4.18 3.79 4.08 F146:CZ F379:CD2 4.38 F167:CZ L404:CD2 3.95
Helix 7 and bL E192:CD H390:CE1 3.59 3.83 3.50 E174:CD H368:CE1 3.74 E196:CD H393:CE1 4.39
Helix 7 and C-terminus L193:CD1 V410:CG1 4.35 4.35 4.54 L175:CD1 V388:CG1 4.11 I197:CD1 V413:CG1 3.55
Side-chain H bonds and electrostatic interactions
bA and helix 1 D23:OD2 R38:NE 2.99 3.23 2.80 D21:OD2 R36:NE2 4.01 D24:OD2 R39:NE 3.38
Helix 1 and bL R38:NE
a
T393:O 5.81 4.04 5.95 R36:NE T371:O 5.57 R39:NE T396:O 2.88
R38:NH2
b
T393:O 5.16 2.68 5.17 R36:NH2 T371:O 4.76 R39:NH2 T396:O 2.93
Helix 1 and bE S45:OG D163:OD2 2.51 2.67 2.60 T43:OG1 D144:OD1 2.64 T46:OG1 D165:OD2 2.73
Helix 1 and bF S45:OG F187:N 2.88 2.98 2.92 T43:OG1 F169:N 2.97 T46:OG1 Y191:N 2.87
Loop after bB and helix 8 R65:NH2

a
D218:OD2 16.25 12.58 14.89 R62:OD1 D200:NH1 16.54 R65:NH1 D222:OD1 3.14
bE and helix 7 (N-term.) D163:OD1 L188:N 2.96 2.88 2.86 D144:OD2 L170:N 2.88 D165:OD1 L192:N 2.84
D163:OD1 M189:N 3.04 2.87 2.89 D144:OD2 M171:N 3.00 D165:OD1 M193:N 2.88
Helix 5 and helix 14 H169:ND1 E400:OE2 3.76 3.77 4.50 H150:ND1 E378:OE2 3.74 H171:ND1 E403:OE1 3.20
Helix 7 and bL E192:OE1 H390:NE2 2.99 2.89 2.82 E174:OE1 H368:NE2 2.90 E196:OE1 H393:NE2 3.06
E192:OE1 S392:OG 2.74 2.92 2.80 E174:OE1 S370:OG 2.72 E196:OE1 S395:OG 2.64
E192:OE2 T393:N 2.63 3.03 2.69 E174:OE2 T371:N 2.95 E196:OE2 T396:N 2.86
E192:OE2 T393:OG1 2.75 2.95 2.56 E174:OE2 T371:OG1 2.79 E196:OE2 T396:OG1 2.48
Helix 8 and bJ K219:NZ
b
N336:OD1 4.65 6.64 6.77 K201:NZ N316:OD1 2.81 K223:NZ N338:OD1 3.23
Substrate-assisted domain–domain cooperativity A. Varga et al.
1876 FEBS Journal 272 (2005) 1867–1885 ª 2005 FEBS
Different stabilizing effects of MgATP and MgADP
are due to their different binding
Different effects of the two nucleotides can be rational-
ized by their different binding modes to PGK, sugges-
ted from thiol-reactivity studies [13], phosphorescence
life-time measurements [28] and supported by X-ray
crystallography [9,83]. Here we present further insight
into these differences. Two crystal structures have been
published with bound MgADP: a binary complex of
the bacterial B. stearothermophilus PGK [9], and a
ternary complex of T. brucei PGK containing bound
MgADP and 3-PG [79]. In spite of their different nat-
ures, the molecular details of MgADP binding are very
similar in these complexes. Only one binary complex
crystal structure exists with bound MgATP [83], but
there are several structures with bound MgATP ana-

logues [4,80,81,84].
The locations and interactions of the adenine and
ribose rings in the C domain are very similar in all
cases. The occupation and interactions of the nucleo-
tide phosphate chain, however, are strikingly differ-
ent, in agreement with solution binding data [83].
Fluctuation of the flexible phosphate chain of
MgATP between two alternative sites was assumed,
while no such argument was required by the well-
positioned phosphates of MgADP [83]. The existence
of two alternative sites for the phosphates of the tri-
nucleotide has, indeed, been shown by the crystallo-
graphic studies with different MgATP analogues
[4,80,81,84].
The binding of MgADP and MgATP are illustrated
by the structures of B. stearothermophilus [9] and pig
muscle [83] PGKs, respectively, in Fig. 5B and C. The
common characteristics of their binding are H-bonding
and hydrophobic interactions of the adenine ring to
bq, and of the ribose ring to the loop between bJ and
helix 12, as well as the ionic interaction of the a-phos-
phate with Lys Bs201 ⁄ 219(NZ) in Bs ⁄ pig muscle PGK
(Fig. 5B and C) at the N terminus of helix 8. Thus,
these secondary structural elements may be similarly
fixed together in the C domain either by MgADP or
MgATP.
There are, however, further interactions, that are dif-
ferent for the two nucleotides. In contrast to MgATP
[83], for MgADP [9] both the a- and b-phosphates are
linked through the bound Mg

2+
to the carboxylate of
an aspartate residue (Asp Bs352 ⁄ 374 in Fig. 5B; the
second number refers to pig muscle PGK) located in
helix 13. The b-phosphate of MgADP is linked to Ser
Bs353 ⁄ Thr375(OG) from the same helix (Fig. 5B).
These multiple interactions of MgADP with helix 13
are further strengthened by H bonds and electrostatic
forces between its positively charged N terminus and
the b-phosphate of the nucleotide. The above inter-
actions result in complete ordering of the helix as
observed by the X-ray data [9]. In addition, a b-phos-
phate O-atom of MgADP is H bonded to the ND2
atom of Asn Bs316 ⁄ 336 from b-strand J (Fig. 5B).
Neither of these interactions is formed with MgATP
(Fig. 5C), therefore helix 13 is not completely ordered
in this binary complex structure.
In addition to the observations noted in the crystal-
lographic papers, here we point out important differ-
ences in formation of contacts between conserved side
chains. In the structure of B. stearothermophilus PGK
with bound MgADP (Fig. 5B) Asn Bs316 ⁄ 336 makes
new H bonds. These are formed between Asn
Bs316 ⁄ 336(OD1) and Lys Bs201 ⁄ 219(NZ) as well as
between Asn Bs316 ⁄ 336(ND2) and Gly Bs349 ⁄ 371(O)
Table 3. (Continued).
Interacting
structural
elements
Pig muscle PGK (open) B. stearotherm. PGK (open) T. brucei PGK (closed)

Atom 1 Atom 2
Substrate
free
3-PG
binary
MgATP
binary Atom 1 Atom 2
MgADP
binary Atom 1 Atom 2 Ternary
c
Helix 8 and helix 14 N225:ND2 L402:O 3.44 2.70 2.71 N207:ND2 M380:O 2.99 N229:ND2 L405:O 6.31
N225:ND2
a
E403:OE1 8.92 8.77 8.64 N207:ND2 E381:OE1 9.08 N229:ND2 E406:OE1 3.51
bJandbK N336:ND2
b
G371:O 6.55 4.21 4.90 N316:ND2 G349:O 3.04 N338:ND2 G374:O 2.72
bJ and helix 13 G337:N
b
T375:OG1 4.75 3.34 3.03 G317:N S353:OG 2.67 G339:N S378:OG 2.93
Helix 12 and loop after bJ T351:OG1
b
V339:N 4.15 3.18 3.39 T331:OG1 M319:N 2.86 T353:OG1 M341:N 3.00
bK and helix 13 G371:O
a
T375:OG1 3.56 5.42 3.59 G349:O S353:OG 3.68 G374:O S378:OG 3.03
Within bL S392:OG T393:N 3.51 3.10 3.26 S370:OG T371:N 3.09 S395:OG T396:N 3.01
S392:OG
b
G394:N 4.03 2.97 3.58 S370:OG G372:N 3.42 S395:OG G397:N 2.84

G394:N T393:OG1 3.30 3.37 3.52 G372:N T371:OG1 3.31 G397:N T396:OG1 3.42
bL and helix 14 S392:OG
b
S398:OG 4.38 4.58 4.26 S370:OG S376:OG 2.72 S395:OG S401:OG 3.00
a
Contacts formed only in the closed structure of T. brucei PGK [79];
b
Contacts formed only in the presence of either substrate.
c
MgADP*3-PG ternary complex.
A. Varga et al. Substrate-assisted domain–domain cooperativity
FEBS Journal 272 (2005) 1867–1885 ª 2005 FEBS 1877
(Fig. 5B). As indicated by the large atomic distances
in Table 3, no such interactions exist in the binary
complex with MgATP (Fig. 5C). Thus, Asn
Bs316 ⁄ 336 from b-strand J has as many as three
interactions including the one with b-phosphate of
MgADP and thereby it connects bJ with the N termi-
nus of helix 8 (Lys Bs201 ⁄ 219) and with the loop
between bK and the N terminus of helix 13 (Gly
Bs349 ⁄ 371). These bonds are characteristic only to
the MgADP complexes [3,9] and are most probably
responsible for the larger conformational stability of
PGK in the presence of MgADP compared to that of
the MgATP complex, as indicated by our calorimetric
measurements.
The interactions above are not restricted to the C
domain. Although it is not seen how the effect of the
nucleotide binding can reach the N domain, the per-
manent contacts of the interdomain region assure

transmission of its effect. In fact, there is a notable
new contact of Ser Bs370 ⁄ 392(OG) of bL with
Bs376 ⁄ 398(OG) at the beginning of helix 14 (belonging
to the N domain) in the presence of MgADP
(Fig. 5B). This contact does not exist in the MgATP
complex (Fig. 5C and Table 3), the phosphate chain of
which occupies an intermediate position between its
two possible alternative sites [83]. These observations
are in line with the cooperative unfolding of the two
domains in the binary complexes with the nucleotides
and with the smaller protecting effect of MgATP than
MgADP.
Highest stability of the active ternary complex
is due to domain closure
Simultaneous protection by the two substrates
against thermal unfolding
The thermal stability of PGK was further increased in
the ternary complexes, i.e. in the simultaneous presence
of two substrates (Fig. 1B and C and Table 1). In
order to avoid the heat effect due to the enzyme reac-
tion, an unproductive ternary complex was investigated
in which instead of the substrate, MgATP, the prod-
uct, MgADP was present. After analysing the data,
similarly to the binary complexes, the activation
parameters (Table 2) were estimated. These correlate
well with the increased stability of the ternary complex
vs. the binary ones.
This increased stability of the ternary complex may
simply be due to simultaneous formation of the char-
acteristic contacts formed in the presence of the indi-

vidual substrates. Thus, even without further
conformational changes (such as domain closure), the
stabilizing effects of the two bound substrates may be
additive. There are, however, convincing arguments
and experimental evidence in favour of PGK domain
closure occurring concomitantly with the enzyme–
substrate complex formation. Since from the above
calorimetric experiments one cannot distinguish
between the conformational stabilities of the open
and closed conformations, in the further work we
followed two different strategies: (a) by analysing the
atomic coordinates of two different (open and closed)
PGK crystal structures we searched for formation of
additional atomic interactions that may be responsible
for stabilization of the closed conformation over the
open one; (b) we devised further calorimetric experi-
ments with a modified PGK that lacks the ability of
domain closure.
Structural origin of the increased stability
of the closed conformation
In the structures of the two closed ternary complexes
of Thermotoga maritima [4] and T. brucei PGK [3] the
characteristic contacts formed upon the separate bind-
ing of 3-PG or MgADP co-exist. The main features of
the contact pattern are similar in the two closed crystal
structures and therefore we discuss only that of the
fully closed T. brucei PGK (Fig. 6).
In detail, on the one hand, the bonds between Arg
Tb39 ⁄ Bs36 ⁄ 38(NH2) and Thr Tb396 ⁄ Bs371 ⁄ 393(O)
(the last numbering refers to pig muscle PGK) as

well as between Gly Tb397 ⁄ Bs372 ⁄ 394(N) and Ser
Tb395 ⁄ Bs370 ⁄ 392(OG), characteristic of the 3-PG bin-
ary complex, also exist in the closed structure of T. bru-
cei PGK. On the other hand, the bond between Ser
Tb395 ⁄ Bs370 ⁄ 392(OG) and Ser Tb401 ⁄ Bs376 ⁄ 398(OG)
as well as the double interactions of Asn Tb338 ⁄
Bs316 ⁄ 336 with Lys Tb223 ⁄ Bs201 ⁄ 219(NZ) and with
Gly Tb374 ⁄ Bs349 ⁄ 371(O), characteristic of the
MgADP binary complex, are also formed in the closed
ternary complex (Fig. 6 and Table 3).
Additional atomic contacts (coloured violet in
Fig. 6), however, are also present in the closed tern-
ary complex structure of T. brucei PGK compared to
those existing separately in the two binary complexes.
The contacts that are different in the closed ternary
and the open binary complexes are highlighted in
Table 3. The contact between Arg Tb39 ⁄ Bs36 ⁄ 38(NE)
and Thr Tb396 ⁄ Bs371 ⁄ 393(O) is essentially character-
istic of the closed conformation. It further strengthens
the contact of the Arg side chain with bL, already
formed in the binary complex with 3-PG. The forma-
tion of a new H bond in the closed structure of
T. brucei PGK, immediately before helix 13, that con-
nects Gly Tb374 ⁄ Bs349 ⁄ 371(O) to Ser ⁄ Thr Tb378 ⁄
Substrate-assisted domain–domain cooperativity A. Varga et al.
1878 FEBS Journal 272 (2005) 1867–1885 ª 2005 FEBS
Bs353 ⁄ 375(OG), stabilizes a new conformation of the
N terminus of helix 13. Another peptide H bond
between the atoms of Gly Tb375 ⁄ Bs350 ⁄ 372(N) and
Ser Tb395 ⁄ Bs370 ⁄ 392(O) (as well as between Gly

Tb375 ⁄ Bs350 ⁄ 372(N) and Gly Tb397 ⁄ Bs352 ⁄ 374(O))
of the closed structure creates a new contact between
bK and bL, i.e. the b strands before and after helix
13 in the sequence. As a further consequence of
domain closure, a new salt bridge is also formed
between Arg Tb65 ⁄ Bs62 ⁄ 65 (N domain) and Asp
Tb222 ⁄ Bs200 ⁄ 218 (C domain) that keep the two
domains together (Fig. 6), while the equivalent residues
in the open structure are far away (Table 3). These addi-
tional contacts are, in fact, characteristic of the closed
conformation, since almost all of them are absent in the
open conformation of crystalline pig muscle PGK tern-
ary complex [81], where domain closure was shown to
be prevented by the crystal lattice forces [35].
The increased number of molecular contacts in the
interdomain region of the closed ternary complex,
compared to the open binary ones, suggests a larger
increase of conformational stability than could have
been expected on the basis of simple additive effects of
the two simultaneously bound substrates.
Protecting effect of substrates against thermal
unfolding is not additive for carboxamidomethylated
(CM)-PGK
In order to test experimentally the possible stabiliza-
tion effect of domain closure on PGK conformation,
thermal unfolding experiments with the chemically
modified enzyme were devised. Pig muscle PGK con-
tains two reactive cysteinyl residues in the hinge region
between the two domains, on the outer surface of helix
13 (Fig. 4A). These thiol-groups do not have any

direct role in enzyme activity, yet their chemical modi-
fication with a relatively bulky reagent (such as car-
boxamidomethylation) leads to loss of enzyme activity
[85]. Small angle X-ray scattering data with this inac-
tive CM-PGK provided evidence for steric prevention
of domain closure [34].
DSC runs of CM-PGK show significantly wider
peaks compared to the unmodified enzyme (Fig. 1A
Fig. 6. Details of the molecular contacts stabilizing the closed conformation. The Ca trace (black) of the molecule of T. brucei [79], PGK is
shown together with the blue ball and stick models of 3-PG and MgADP. The Ca traces of the important secondary structural elements are
coloured in the same way as in Figs. 4 and 5. From the side-chain and backbone atoms (stick models) only those are shown (in the colour
of the corresponding secondary structural elements) that either directly interact with the substrates or are involved in important contacts
within the protein and exist already in the respective binary substrate complexes. The side-chain or backbone atoms that participate in newly
formed contacts in the closed structure are coloured violet. The interacting atoms are connected with dashed lines. The contacts within the
protein together with the distances between the corresponding atoms are listed in Table 3.
A. Varga et al. Substrate-assisted domain–domain cooperativity
FEBS Journal 272 (2005) 1867–1885 ª 2005 FEBS 1879
and B), suggesting a less cooperative heat transition.
In addition, there is a remarkable loss of stability of
the whole molecule, as indicated by the large decrease
both in the T
m
values and in the calorimetric heat (Q
t
)
(Table 1). Plotting the data in the way shown in
Fig. 2, can be approached by a straight line (not
shown) and the activation parameters for thermal
unfolding of this modified PGK are clearly different
from those for the intact active enzyme (Table 2). The

experiment also showed that the intact state of helix
13 is important for the stability of the whole molecule.
Previous studies with pig muscle PGK suggest that
introduction of the chemical label perturbs the con-
formational state of helix 13, and thereby may also
disrupt important stabilizing contacts between the two
domains [80].
The most interesting part of the results is the pro-
tective effects of the substrates against thermal unfold-
ing of CM-PGK. Protection by each substrate
correlates well with the full substrate binding ability of
this modified enzyme [13]. The order of effects by each
substrate in the binary complexes is the same as found
for the unmodified active PGK (Fig. 1A and B;
Tables 1 and 2). There is, however, a remarkable dif-
ference between the modified and unmodified enzymes:
the ternary complex of CM-PGK does not exhibit the
highest stability among the enzyme–substrate com-
plexes. This can only be due to lack of domain closure,
as substrate binding to CM-PGK has failed to induce
domain closure as shown in a previous small angle
X-ray scattering study [34].
Thus, our results revealed that the highest confor-
mational stability is a characteristic of the native,
functionally competent ternary complex. The larger
protection by substrates in the ternary complex (vs. the
binary ones) is abolished in the modified inactive
CM-PGK, where domain closure is prevented. The
increased conformational stability of the closed confor-
mation has been also judged by surveying the crystal

structure. Taken together, the calorimetric experiments,
in accordance with the crystal structures, suggest that
the domain-closed conformation is an exclusive charac-
teristic of the ternary complex, i.e. both substrates are
required for inducing domain closure.
Experimental procedures
Enzymes and chemicals
Pig muscle PGK was isolated as described earlier [8] and
stored as a microcrystalline suspension in the presence of
ammonium sulphate and 2 mm dithiothreitol. Yeast PGK
was obtained from Sigma (St. Louis, MO, USA) as a preci-
pitate in the presence of 3 m ammonium sulphate. Their
activity was determined using 3-PG and MgATP as sub-
strates in a coupled assay using GAPDH (prepared from
pig muscle [86]) as auxiliary enzyme and varied between
300 and 500 katÆmol
)1
.
Two mutants of yeast PGK were constructed: the W308F,
W333F, Y122W triple mutant (denoted as W122), and the
W308F point mutant (W333). The mutants contain a single
Trp residue either in the N- (W122), or the C-terminal
domain (W333). The expression and purification of the
mutants were described recently [62]. Both mutants were
His-tagged, and purified using Ni
+
affinity chromatography.
The enzyme activity of the W122 and W333 mutants was
measured after purification. Neither the His-tag, nor the
mutations influenced significantly the catalytic activity.

The sodium salts of 3-PG, ATP, ADP and NADH were
Boehringer products. The complexes of MgATP, MgADP
were formed by adding MgCl
2
(Sigma) in molar excess
(12 mm), which assured a practically complete saturation
on the basis of the dissociation constants of 0.1 and
0.6 mm, respectively, which were averaged from the litera-
ture [87–91]. Iodoacetamide was from Sigma. All other
chemicals were reagent-grade commercial products.
Preparation of enzyme solutions
Crystals of pig muscle PGK or the precipitates of the wild-
type yeast PGK were dissolved in 50 mm Tris ⁄ HCl pH 7.5
containing 1 mm EDTA and 1 mm dithiothreitol and
dialysed against the same buffer to remove (NH
4
)
2
SO
4
.
GAPDH was similarly desalted. Dialysis of the yeast PGK
mutants was not required, as they were stored in desalted
frozen solutions [62]. Protein concentration was determined
from the UV absorption at 280 nm using the method des-
cribed by Pace and coworkers [92]. The values of 27 900,
21 400, 14 400 and 15 900 m
)1
Æcm
)1

were obtained for the
pig PGK, the wild-type yeast PGK, W122 and W333
mutants of yeast PGK, respectively. The molecular mass of
PGK was taken uniformly to be 44.5 kDa.
Alkylation of pig muscle PGK was performed with iodo-
acetamide as described previously [85]. The reaction was
stopped by the addition of 10 mm dithiothreitol and the
excess of the reagents was removed by dialysis.
DSC
DSC measurements were performed on a MicroCal VP-
DSC type microcalorimeter (MicroCal Incorporate, North-
ampton, MA, USA) with a cell volume of 0.51 mL. The
applied scanning rates are given in the Figure legends. The
protein concentration was 0.003 mm (0.13 mgÆmL
)1
) in all
experiments. All samples were carefully degassed before the
experiments. The results were analysed using microcal
origin 5.0 software. The melting temperature (T
m
) was
Substrate-assisted domain–domain cooperativity A. Varga et al.
1880 FEBS Journal 272 (2005) 1867–1885 ª 2005 FEBS
determined after subtraction of the instrumental baseline.
In the heat induced unfolding experiments carried out in
the presence of substrates, high concentrations of ligands
(much above the normal saturation) were used. It was
checked that further increase of ligand concentrations did
not affect the results.
Fluorimetric experiments

Fluorescence detected heat induced unfolding was carried
out using a SPEX Fluoromax-3 spectrofluorometer
equipped with a Peltier thermostat (Edison, NJ, USA). The
same heating rate and protein concentration was used as
for the DSC. Fluorescence was excited at 295 nm and mon-
itored at its emission maximum using 2 nm bandwidths on
both sides with a 1 cm path length for excitation and 4 mm
for emission. While the temperature was increased at a rate
of 1°Æmin
)1
, the fluorescence intensity was recorded at every
0.33 min. The measured fluorescence intensities were con-
verted into the apparent fraction of native protein, F
N
, vs.
temperature according to the Eqn (1) [93].
F
N
¼
ðI
U
þ m
U
TÞÀI
ðI
U
þ m
U
TÞÀðI
N

þ m
N
TÞ
ð1Þ
where I is the fluorescence intensity measured at tempera-
ture T, and I
N
and I
U
are the intercepts and m
N
and m
U
are the slopes of the pre- and post-transitional base lines of
the raw data, respectively.
Quantitative analysis of the experimental data
Thermal unfolding experiment were analysed according to
Sanchez-Ruiz et al. [77], by assuming the two-state irrevers-
ible model:
N À!
k
D ð2Þ
where N and D are the native and the denatured forms of
the protein and k is the first order rate constant.
It was shown that the first order rate constant (k) of this
process is related to the shape of the calorimetrically deter-
mined heat capacity curve in the following way [77]:
k ¼
vC
p

Q
t
À Q
ð3Þ
where v is the scanning rate, C
p
is the experimental heat
capacity, Q
t
is the total heat of the process (identical to the
integrated area under the transition peak) and Q is the heat
evolved at a given temperature.
The rate constants obtained in this way were analysed as
a function of the temperature according to the Arrhenius
relationship:
k ¼ A
expðÀE=RTÞ
ð4Þ
where A is the pre-exponential factor, E is the activation
energy, R is the gas constant and T is the absolute tempera-
ture. The terms A and E can be expressed in termodynamic
quantities:
A ¼
k
B
T
h
and E ¼ DG
z
ð5Þ and ð6Þ

where k
B
is the Boltzmann constant, h is Planck constant
and DG
à
is the activation free enthalpy of the unfolding
process. Although this expression of the pre-exponential
factor is rigorously valid only in the gas phase and neglects
the solvent effects, we apply this relationship here as an
approximation that affects the absolute maxima of the acti-
vation barrier, but probably has little effect on the relative
results for the different proteins ⁄ mutants. To obtain the
activation enthalpy (DH
à
) and the activation entropy (DS
à
)
of the process separately, the Arrhenius equation can be
transformed accordingly:
ln
k
T
¼ ln
k
B
h
þ
DS
z
R

À
DH
z
RT
ð7Þ
The plot of ln(k ⁄ T) vs. 1 ⁄ T gives a straight line, the slope
and the intercept on the ordinate give DH
à
and DS
à
,
respectively.
The calorimetric data were also analysed according to
another type of transformation of the data, as described by
Sanchez-Ruiz et al. [77]:
ln ln
Q
t
Q
t
À Q

¼
E
RT
m
À
E
RT
ð8Þ

where T
m
is the transition temperature. The plot of
ln[ln(Q
t
⁄ Q
t
–Q)] vs. 1 ⁄ T gives a straight line.
The same equation can be used to analyse the fluorimet-
rically detected heat transition, since the following relation-
ship holds:
F
N
¼
½N
0
½N
¼
Q
t
Q
t
À Q
ð9Þ
where F
N
is the fraction of the native protein, as defined by
Eqn (1), [N]
0
and [N] are the molar fractions of the native

state, initially and at any given temperature, respectively.
Thus, from the fluorimetric data the plot of ln(ln([N]
0
⁄ [N])
against 1 ⁄ T was prepared, on the basis of Eqn (8).
Computer molecular graphics
For visualizing and analysing the molecular details of the 12
PGK structures studied the insight ii (Biosym ⁄ MSI, San
Diego, CA, USA) software was used. The pdb codes of the
protein co-ordinates used in structural comparisons are as
follows: 1PHP (B. stearothermophilus PGK*MgADP [9]),
1VJC (pig muscle PGK*MgATP [83]), 1HDI (pig muscle
PGK*3PG*MgADP [5]), 1KF0 (pig muscle PGK*3PG*
MgAMPPCP [80]), 13PK (T. brucei PGK*3PG*MgADP
[79]), 1VPE (T. maritima PGK*3PG*MgAMPPNP [4]),
3PGK (substrate-free yeast PGK [7]) and 1QPG (mutant
yeast PGK*3PG*MgAMPPNP [84]). The pdb co-ordinates
A. Varga et al. Substrate-assisted domain–domain cooperativity
FEBS Journal 272 (2005) 1867–1885 ª 2005 FEBS 1881
of pig muscle PGK*3PG binary [8], pig muscle
PGK*3PG*MnAMPPNP [81], yeast PGK*3-PG binary
complex [82] and of the substrate-free pig muscle PGK [80]
were obtained from the authors. Mapping of the molecular
contacts responsible for maintaining the three-dimensional
structure of PGK was carried out in three steps. First, the
conserved residues were selected after making alignments of
168 PGK sequences found in ExPASy Molecular Biology
Server. As a second, the lists of all peptide H bonds, the
hydrophobic, electrostatic and H-bonding contacts of the
conserved side chains (besides the contacts between two side

chains the contacts including a side chain atom with a back-
bone atom are also considered) were enumerated separately
for all the investigated structures. The distance limits for H
bonds, hydrophobic and ionic interactions were 3.5, 4.5 and
4.0 A
˚
, respectively. In one case, from all type of contacts only
those ones were selected which invariantly exist in all 12
structures. In the other case, only the variable contacts were
considered, characteristic of each of the substrate complexes.
In the third step the selected contacts were visualized and
compared for related pairs of structures superposed accord-
ing to the core b strands of either the N- or the C-terminal
domain. PGK structures from different organisms were com-
pared using the structure based sequence alignment, prepared
by us [5] and completed here with the data of yeast and Bs
PGKs.
Acknowledgements
Thanks are due to K. Harlos and A. May (Laboratory
of Molecular Biophysics, University Oxford, UK),
N.R. Chandra (Molecular Biophysics Unit, Indian
Institute of Sciences, Bangalore, Karnataka, India)
and to Z. Kova
´
ri (Institute of Enzymology, BRC,
Hung. Acad. Sci., Budapest, Hungary) for providing
the unpublished pdb co-ordinates of pig muscle
PGK*3PG binary, pig muscle PGK*3PG*MnAMP-
PNP, yeast PGK*3-PG binary complex and of the
substrate-free pig muscle PGK, respectively. Financial

support by grants OTKA (F 038395 and T 043446) of
the Hungarian National Research Fund are gratefully
acknowledged. B.F and S.O. were funded by the grants
OTKA D 48578 and D 38480, respectively.
References
1 Daniel RM, Dunn RV, Finney JL & Smith JC (2003)
The role of dynamics in enzyme activity. Annu Rev Bio-
phys Biomol Struct 32, 69–92.
2 Banks RD, Blake CCF, Evans PR, Haser R, Rice
DW, Hardy GW, Merrett M & Phillips AW (1979)
Sequence, structure and activity of phosphoglycerate
kinase: a possible hinge-bending enzyme. Nature 279,
773–777.
3 Bernstein BE, Michels PAM & Hol WGJ (1997) Synergis-
tic effects of substrate-induced conformational changes in
phosphoglycerate kinase activation. Nature 385, 275–278.
4 Auerbach G, Huber R, Gra
¨
ttinger M, Zaiss K, Schurig
H, Jaenicke R & Jacob U (1997) Closed structure of
phosphoglycerate kinase from Thermotoga maritima
reveals the catalytic mechanism and determinants of
thermal stability. Structure 5, 1475–1483.
5 Szila
´
gyi AN, Ghosh M, Garman E & Vas M (2001) A
1.8 A resolution structure of pig muscle 3-phosphoglyce-
rate kinase with bound MgADP and 3-phosphoglycerate
in open conformation: new insight into the role of the
nucleotide in domain closure. J Mol Biol 306, 499–511.

6 Mori N, Singer-Sam J & Riggs AD (1986) Evolutionary
conservation of the substrate binding cleft of 3-phos-
phoglycerate kinases. FEBS Lett 204, 313–317.
7 Watson HC, Walker NPC, Shaw PJ, Bryant TN,
Wendell PL, Fothergill L, Perkin RE, Conroy SC,
Dobson MJ, Tuite MF, Kingsman AJ & Kingsman
SM (1982) Sequence and structure of yeast phospho-
glycerate kinase. EMBO J 1, 1635–1640.
8 Harlos K, Vas M & Blake CCF (1992) Crystal structure
of the binary complex of pig muscle phosphoglycerate
kinase and its substrate 3-phospho-D-glycerate. Proteins
12, 133–144.
9 Davies GJ, Gamblin SJ, Littlechild JA, Dauter Z, Wilson
KS & Watson HC (1994) Structure of the ADP complex
of the 3-phosphoglycerate kinase from Bacillus stearo-
thermophilus at 1.65 A. Acta Crystallogr D50, 202–209.
10 Hjelmgren T, Strid L & Arvidsson L (1976) An essential
arginyl residue in phosphoglycerate kinase from yeast.
FEBS Lett 68, 137–140.
11 Desvages G, Roustan C, Fattoum A & Pradel LA
(1980) Structural studies on yeast 3-phosphoglycerate
kinase. Identification by immuno-affinity chromatogra-
phy of one glutamyl residue essential for yeast 3-phos-
phoglycerate kinase activity. Its location in the primary
structure. Eur J Biochem 105, 259–266.
12 Cserpa
´
n I & Vas M (1983) Effects of substrates on the
heat stability and on the reactivities of thiol groups of
3-phosphoglycerate kinase. Eur J Biochem 131, 157–162.

13 Tompa P, Hong PT & Vas M (1986) The phosphate
group of 3-phosphoglycerate accounts for conforma-
tional changes occurring on binding to 3-phosphoglyce-
rate kinase. Enzyme inhibition and thiol reactivity
studies. Eur J Biochem 154, 643–649.
14 Mas MT, Resplandor ZE & Riggs AD (1987) Site-direc-
ted mutagenesis of glutamate-190 in the hinge region of
yeast 3-phosphoglycerate kinase: implications for the
mechanism of domain movement. Biochemistry 26,
5369–5377.
15 Desmadril M, Minard P, Ballery N, Gaillard-Miran S,
Hall L & Yon JM (1991) Conformational changes in
yeast phosphoglycerate kinase upon ligand binding:
Substrate-assisted domain–domain cooperativity A. Varga et al.
1882 FEBS Journal 272 (2005) 1867–1885 ª 2005 FEBS
fluorescence of a linked probe and chemical reactivity of
genetically introduced cysteinyl residues. Proteins 10,
315–324.
16 Wrobel JA & Stinson RA (1981) The effects of anions,
substrates, metal ions and sulfhydryl reagents on the
proteolytic susceptibility of yeast phosphoglycerate
kinase, Biochim Biophys Acta 662, 236–245.
17 Jiang SX & Vas M (1988) Limited proteolysis of
3-phosphoglycerate kinase without loss of enzymic
activity. FEBS Lett 231, 151–154.
18 Adams B & Pain RH (1986) The effect of substrates
on the inter–domain interactions of the hinge-bending
enzyme 3-phosphoglycerate kinase. FEBS Lett 196,
361–364.
19 Hu CQ & Sturtevant JM (1987) Thermodynamic study of

yeast phosphoglycerate kinase. Biochemistry 26, 178–182.
20 Brandts JF, Hu CQ, Lin LN & Mas MT (1989) A sim-
ple model for proteins with interacting domains. Appli-
cations to scanning calorimetry data. Biochemistry 28,
8588–8596.
21 Tanswell P, Westhead EW & Williams RJP (1976)
Nuclear-magnetic-resonance study of the active-site
structure of yeast phosphoglycerate kinase. Eur J Bio-
chem 63, 249–262.
22 Wilson HR, Williams RJP, Littlechild JA & Watson
HC (1988) NMR analysis of the interdomain region of
yeast phosphoglycerate kinase. Eur J Biochem 170, 529–
538.
23 Graham HC, Williams RJ, Littlechild JA & Watson HC
(1991) A proton-NMR study of a site-directed mutation
(His388Glu) in the interdomain region of yeast phos-
phoglycerate kinase. Implications for domain move-
ment. Eur J Biochem 196 , 261–269.
24 Wasylewski Z & Eftink MR (1987) Frequency domain
fluorescence studies of yeast phosphoglycerate kinase
and its ternary complex. Eur J Biochem 167, 513–518.
25 Mouawad L, Desmadril M, Perahia D, Yon JM & Bro-
chon JC (1990) The effects of ligands on the conforma-
tion of phosphoglycerate kinase: fluorescence anisotropy
decay and theoretical interpretation. Biopolymers 30,
1151–1160.
26 Haran G, Haas E, Szpikowska BK & Mas MT (1992)
Domain motions in phosphoglycerate kinase: determina-
tion of interdomain distance distributions by site-specific
labeling and time-resolved fluorescence energy transfer.

Proc Natl Acad Sci USA 89, 11764–11768.
27 Dryden DT, Varley PG & Pain RH (1992) A study of
the hinge-bending mechanism of yeast 3-phosphoglyce-
rate kinase. Eur J Biochem 208, 115–123.
28 Cioni P, Puntoni A & Strambini GB (1993) Tryptophan
phosphorescence as a monitor of the solution structure
of phosphoglycerate kinase from yeast. Biophys Chem
46, 47–55.
29 Cheung CW & Mas MT (1996) Substrate-induced con-
formational changes in yeast 3-phosphoglycerate kinase
monitored by fluorescence of single tryptophan probes.
Protein Sci 5, 1144–1149.
30 Vlasova II & Kuprin SP (1996) An electron parameg-
netic resonance study of phosphoglycerate kinase from
yeast. I. Modification of protein SH-groups by spin-
labels, conformational changes of the enzyme., Biofizika
(Russian). 41, 1193–1200.
31 Roustan C, Fattoum A, Jeanneau R & Pradel LA
(1980) Yeast 3-phosphoglycerate kinase: sulfate and
substrate binding, their effect on the conformational
state of the enzyme. Biochemistry 19, 5168–5175.
32 Pickover CA, McKay DB, Engelman DM & Steitz TA
(1979) Substrate binding closes the cleft between the
domains of yeast phosphoglycerate kinase. J Biol Chem
254, 11323–11329.
33 Ptitsyn OB, Pavlov MY, Sinev MA & Timchenko AA
(1986) Study of Domain Displacements in Proteins by
Diffuse X-Ray Scattering. Akade
´
miai Kiado

´
, Budapest.
34 Sinev MA, Razgulyaev OI, Vas M, Timchenko AA &
Ptitsyn OB (1989) Correlation between enzyme activity
and hinge-bending domain displacement in 3-phospho-
glycerate kinase. Eur J Biochem 180, 61–66.
35 Kova
´
ri Z & Vas M (2004) Protein conformer selection
by sequence-dependent packing contacts in crystals of
3-phosphoglycerate kinase. Proteins 55, 198–209.
36 Betton JM, Desmadril M, Mitraki A & Yon JM (1984)
Unfolding-refolding transition of a hinge bending
enzyme: horse muscle phosphoglycerate kinase induced
by guanidine hydrochloride. Biochemistry 23, 6654–6661.
37 Desmadril M, Mitraki A, Betton JM & Yon JM (1984)
GuHCl induced unfolding-folding transition of a hinge-
bending protein: horse muscle phosphoglycerate kinase.
Biochem Biophys Res Commun 118, 416–422.
38 Missiakas D, Betton JM, Minard P & Yon JM (1990)
Unfolding-refolding of the domains in yeast phospho-
glycerate kinase: comparison with the isolated engi-
neered domains. Biochemistry 29, 8683–8689.
39 Szpikowska BK, Beechem JM, Sherman MA & Mas
MT (1994) Equilibrium unfolding of yeast phosphogly-
cerate kinase and its mutants lacking one or both native
tryptophans: a circular dichroism and steady-state and
time-resolved fluorescence study. Biochemistry 33, 2217–
2225.
40 Szpikowska BK & Mas MT (1996) Urea-induced equili-

brium unfolding of single tryptophan mutants of yeast
phosphoglycerate kinase: evidence for a stable inter-
mediate. Arch Biochem Biophys 335, 173–182.
41 Sherman MA, Beechem JM & Mas MT (1995) Probing
intradomain and interdomain conformational changes
during equilibrium unfolding of phosphoglycerate
kinase: fluorescence and circular dichroism study of
tryptophan mutants. Biochemistry 34, 13934–13942.
42 Adams B, Burgess RJ & Pain RH (1985) The folding
and mutual interaction of the domains of yeast 3-phos-
phoglycerate kinase. Eur J Biochem 152, 715–720.
A. Varga et al. Substrate-assisted domain–domain cooperativity
FEBS Journal 272 (2005) 1867–1885 ª 2005 FEBS 1883
43 Beechem JM, Sherman MA & Mas MT (1995) Sequen-
tial domain unfolding in phosphoglycerate kinase: fluor-
escence intensity and anisotropy stopped-flow kinetics
of several tryptophan mutants. Biochemistry 34, 13943–
13948.
44 Lillo MP, Szpikowska BK, Mas MT, Sutin JD & Bee-
chem JM (1997) Real-time measurement of multiple
intramolecular distances during protein folding reac-
tions: a multisite stopped-flow fluorescence energy-trans-
fer study of yeast phosphoglycerate kinase. Biochemistry
36, 11273–11281.
45 Betton JM, Desmadril M & Yon JM (1989) Detection
of intermediates in the unfolding transition of phospho-
glycerate kinase using limited proteolysis. Biochemistry
28, 5421–5428.
46 Ballery N, Desmadril M, Minard P & Yon JM (1993)
Characterization of an intermediate in the folding path-

way of phosphoglycerate kinase: Chemical reactivity of
genetically introduced cysteinyl residues during the fold-
ing process. Biochemistry 32, 708–714.
47 Parker MJ, Spencer J, Jackson GS, Burston SG, Hosszu
LLP, Craven CJ, Waltho P & Clarke AR (1996)
Domain behaviour during refolding of a thermostable
phosphoglycerate kinase. Biochemistry 35, 15740–15752.
48 Betton JM, Desmadril M, Mitraki A & Yon JM (1985)
Kinetic studies of the unfolding-refolding of horse mus-
cle phosphoglycerate kinase induced by guanidine
hydrochloride. Biochemistry 24, 4570–4577.
49 Mitraki A, Betton JM, Desmadril M & Yon JM (1987)
Quasi-irreversibility in the unfolding-refolding transition
of phosphoglycerate kinase induced by guanidine hydro-
chloride. Eur J Biochem 163, 29–34.
50 Hosszu´ LLP, Craven CJ, Parker MJ, Lorch M, Spencer
J, Clarke AR & Waltho JP (1997) Structure of a kinetic
protein folding intermediate by equlibrium amide
exchange. Nature Struct Biol 4, 801–804.
51 Pecorari F, Minard P, Desmadril M & Yon JM (1993)
Structure and functional complementation of engineered
fragments from yeast phosphoglycerate kinase. Protein
Eng 6, 313–325.
52 Pecorari F, Guilbert C, Minard P, Desmadril M & Yon
JM (1996) Folding and functional complementation of
engineered fragments from yeast phosphoglycerate
kinase. Biochemistry 35, 3465–3476.
53 Minard P, Hall L, Betton JM, Missiakas D & Yon JM
(1989) Efficient expression and characterization of iso-
lated structural domains of yeast phosphoglycerate

kinase generated by site-directed mutagenesis. Protein
Eng 3, 55–60.
54 Fairbrother WJ, Minard P, Hall L, Betton JM, Missia-
kas D, Yon JM & Williams RJ (1989) Nuclear magnetic
resonance studies of isolated structural domains of yeast
phosphoglycerate kinase. Protein Eng 3, 5–11.
55 Sherman MA, Chen Y & Mas MT (1997) An engi-
neered amino-terminal domain of yeast phosphoglyce-
rate kinase with native-like structure. Protein Sci 6,
882–891.
56 Betton JM, Missiakas D & Yon JM (1992) The
slow-refolding step of phosphoglycerate kinase as moni-
tored by pulse proteolysis. Arch Biochem Biophys 296,
95–101.
57 Vas M, Sinev MA, Kotova N & Semisotnov GV (1990)
Reactivation of 3-phosphoglycerate kinase from its pro-
teolytic fragments. Eur J Biochem 189, 575–579.
58 Semisotnov GV, Vas M, Chemeris VV, Kashparova NJ,
Kotova NV, Razgulyaev OI & Sinev MA (1991) Refold-
ing of pig muscle and yeast 3-phosphoglycerate kinases
and of their proteolytic fragments. Eur J Biochem 202,
1083–1089.
59 Missiakas D, Betton JM, Chaffotte A, Minard P & Yon
JM (1992) Kinetic studies of the refolding of yeast phos-
phoglycerate kinase: comparison with the isolated engi-
neered domains. Protein Sci 1, 1485.
60 Szila
´
gyi AN & Vas M (1998) Sequential domain refold-
ing of pig muscle 3-phosphoglycerate kinase: kinetic

analysis of reactivation. Fold Des 3, 565–575.
61 Szila
´
gyi AN, Kotova NV, Semisotnov GV & Vas M
(2001) Incomplete refolding of a fragment of the
N-terminal domain of pig muscle 3-phosphoglycerate
kinase that lacks a subdomain: Comparison with refold-
ing of the complementary C-terminal fragment. Eur J
Biochem 268, 1851–1860.
62 Osva
´
th S, Sabelko JJ & Gruebele M (2003) Tuning the
heterogeneous early folding dynamics of phosphoglyce-
rate kinase. J Mol Biol 333, 187–199.
63 Hosszu´ LLP, Craven CJ, Spencer J, Parker MJ, Clarke
AR, Kelly M & Waltho JP (1997) Is the structure of the
N-domain of phosphoglycerate kinase affected by isola-
tion from the intact molecule? Biochemistry 36, 333–340.
64 Nojima H, Ikai A, Oshima T & Noda H (1977) Reversi-
ble thermal unfolding of thermostable phosphoglycerate
kinase. Thermostability associated with mean zero
enthalpy change. J Mol Biol 116, 429–442.
65 Nojima H & Noda H (1979) Kinetics of thermal unfold-
ing and refolding of thermostable phosphoglycerate
kinase. J Biochem (Tokyo) 86 , 1055–1065.
66 Griko YV, Venyaminov SY & Privalov PL (1989) Heat
and cold denaturation of phosphoglycerate kinase
(interaction of domains). FEBS Lett 244, 276–278.
67 Galisteo ML, Mateo PL & Sanchez-Ruiz JM (1991)
Kinetic study on the irreversible thermal denaturation

of yeast phosphoglycerate kinase. Biochemistry 30,
2061–2066.
68 Johnson CM, Cooper A & Brown AJ (1991) A compar-
ison of the reactivity and stability of wild-type and
His388Gln mutant phosphoglycerate kinase from yeast.
Eur J Biochem 202, 1157–1164.
69 Freire E, Murphy KP, Sanchez-Ruiz JM, Galisteo ML
& Privalov PL (1992) The molecular basis of coopera-
tivity in protein folding. Thermodynamic dissection of
Substrate-assisted domain–domain cooperativity A. Varga et al.
1884 FEBS Journal 272 (2005) 1867–1885 ª 2005 FEBS
interdomain interactions in phosphoglycerate kinase.
Biochemistry 31, 250–256.
70 Damaschun G, Damaschun H, Gast K, Misselwitz R,
Muller JJ, Pfeil W & Zirwer D (1993) Cold denatura-
tion-induced conformational changes in phosphoglyce-
rate kinase from yeast. Biochemistry 32, 7739–7746.
71 Gast K, Damaschun G, Damaschun H, Misselwitz R &
Zirwer D (1993) Cold denaturation of yeast phospho-
glycerate kinase: kinetics of changes in secondary struc-
ture and compactness on folding and refolding.
Biochemistry 32, 7747–7752.
72 Gast K, Damaschun G, Desmadril M, Minard P, Mu
¨
l-
ler-Frohne M, Pfeil W & Zirwer D (1995) Cold dena-
turation of yeast phosphoglycerate kinase: which
domain is more stable? FEBS Lett 358, 247–250.
73 Receveur V, Garcia P, Durand D, Vachette P & Desma-
dril M (2000) Role of hydrophobic interactions in yeast

phosphoglycerate kinase stability. Proteins 38, 226–238.
74 Grattinger M, Dankesreiter A, Schurig H & Jaenicke R
(1998) Recombinant phosphoglycerate kinase from the
hyperthermophilic bacterium Thermotoga maritima: cat-
alytic, spectral and thermodynamic properties. J Mol
Biol 280, 525–533.
75 Zaiss K & Jaenicke R (1999) Thermodynamic study of
phosphoglycerate kinase from Thermotoga maritima and
its isolated domains: reversible thermal unfolding moni-
tored by differential scanning calorimetry and circular
dichroism spectroscopy. Biochemistry 38 , 4633–4639.
76 Bentahir M, Feller G, Aittaleb M, Lamotte-Brasseur J,
Himri T, Chessa JP & Gerday C (2000) Structural,
kinetic, and calorimetric characterization of the cold-
active phosphoglycerate kinase from the antarctic Pseu-
domonas sp. TACII18. J Biol Chem 275, 11147–11153.
77 Sanchez-Ruiz JM, Lopez-Lacomba JL, Cortijo M &
Mateo PL (1988) Differential scanning calorimetry of
the irreversible thermal denaturation of thermolysin.
Biochemistry 27, 1648–1652.
78 Kurganov BI, Lyubarev AE, Sanchez-Ruiz JM &
Shnyrov VL (1997) Analysis of differential scanning
calorimetry data for proteins: Criteria of validity of
one-step mechanism of irreversible protein denaturation.
Biophys Chem 69, 125–135.
79 Bernstein BE & Hol WG (1998) Crystal structures of
substrates and products bound to the phosphoglycerate
kinase active site reveal the catalytic mechanism. Bio-
chemistry 37, 4429–4436.
80 Kova

´
ri Z, Flachner B, Na
´
ray-Szabo
´
G & Vas M (2002)
Crystallographic and thiol-reactivity studies on the com-
plex of pig muscle phosphoglycerate kinase with ATP
analogues: correlation between nucleotide binding mode
and helix flexibility. Biochemistry 41, 8796–8806.
81 May A, Vas M, Harlos K & Blake CCF (1996) 2.0 A
resolution structure of a ternary complex of pig muscle
phosphoglycerate kinase containing 3-phospho-D-glyce-
rate and the nucleotide Mn adenylylimidodiphosphate.
Proteins 24, 292–303.
82 Chandra NR, Muirhead H, Holbrook JJ, Bernstein BE,
Hol WG & Sessions RB (1998) A general method of
domain closure is applied to phosphoglycerate kinase
and the result compared with the crystal structure of a
closed conformation of the enzyme. Proteins 30, 372–380.
83 Flachner B, Kova
´
ri Z, Varga A, Gugolya Z, Von-
derviszt F, Na
´
ray-Szabo
´
G & Vas M (2004) Role of
phosphate chain mobility of MgATP in completing the
3-phosphoglycerate kinase catalytic site: binding,

kinetic, and crystallographic studies with ATP and
MgATP. Biochemistry 43, 3436–3449.
84 McPhillips TM, Hsu BT, Sherman MA, Mas MT &
Rees DC (1996) Structure of the R65Q mutant of yeast
3-phosphoglycerate kinase complexed with Mg-AMP-
PNP and 3-phospho-D-glycerate. Biochemistry 35,
4118–4127.
85 De
´
ka
´
ny K & Vas M (1984) Inactivation of pig muscle
3-phosphoglycerate kinase by thiol modification depends
on reagent size. Eur J Biochem 139, 125–130.
86 Elo
¨
di P & Szo
¨
re
´
nyi E (1956) Crystallisation and com-
parative studies of d-glyceraldehyde-3-phosphate dehy-
drogenase from muscle of various mammals. Acta
Physiol Acad Sci Hung 9, 339–350.
87 Burton K (1959) Formation constants for complexes of
adenosine di- or tri-phosphate with magnesium or cal-
cium ions. Biochem J 71, 388–395.
88 Larsson-Raznikiewicz M (1964) Kinetic studies on the
reaction catalysed by phosphoglycerate kinase. I. The
effect of Mg2+ and adenosine-5¢-triphosphate. Biochim

Biophys Acta 85, 60–68.
89 Miller C, Frey CM & Stuehr JE (1972) Interactions of
divalent metal ions with inorganic and nucleotide phos-
phates. I. Thermodynamics. J Am Chem Soc 94, 8898–
8904.
90 Gupta RK, Gupta P, Yashok WP & Rose ZB (1983)
Measurement of the dissociation constant of MgATP at
physiological nucleotide levels by a combination of 31P
NMR and optical absorbance spectroscopy. Biochem
Biophys Res Comm 117, 210–216.
91 Zhang W, Truttmann AC, Luthi D & McGuigan JA
(1997) Apparent Mg
2+
-adenosine 5-triphosphate disso-
ciation constant measured with Mg
2+
macroelectrodes
under conditions pertinent to 31P NMR ionized mag-
nesium determinations. Anal Biochem 251, 246–250.
92 Pace CN, Vajdos F, Fee L, Grimsley G & Gray T
(1995) How to measure and predict the molar absorp-
tion coefficient of a protein. Protein Sci 4, 2411–2423.
93 Wang, C., Lascu, I. & Giartosio, A. (1998) Bovine
serum fetuin is unfolded through a molten globule state,
Biochemistry 37, 8457–8464.
A. Varga et al. Substrate-assisted domain–domain cooperativity
FEBS Journal 272 (2005) 1867–1885 ª 2005 FEBS 1885