Millisecond dynamics in glutaredoxin during catalytic turnover is dependent on substrate binding and absent in the resting states

Article abstract of DOI:10.1021/ja1096539

Conformational dynamics is important for enzyme function. Which motions of enzymes determine catalytic efficiency and whether the same motions are important for all enzymes, however, are not well understood. Here we address conformational dynamics in glutaredoxin during catalytic turnover with a combination of NMR magnetization transfer, R₂ relaxation dispersion, and ligand titration experiments. Glutaredoxins catalyze a glutathione exchange reaction, forming a stable glutathinoylated enzyme intermediate. The equilibrium between the reduced state and the glutathionylated state was biochemically tuned to exchange on the millisecond time scale. The conformational changes of the protein backbone during catalysis were followed by ¹⁵N nuclear spin relaxation dispersion experiments. A conformational transition that is well described by a two-state process with an exchange rate corresponding to the glutathione exchange rate was observed for 23 residues. Binding of reduced glutathione resulted in competitive inhibition of the reduced enzyme having kinetics similar to that of the reaction. This observation couples the motions observed during catalysis directly to substrate binding. Backbone motions on the time scale of catalytic turnover were not observed for the enzyme in the resting states, implying that alternative conformers do not accumulate to significant concentrations. These results infer that the turnover rate in glutaredoxin is governed by formation of a productive enzyme-substrate encounter complex, and that catalysis proceeds by an induced fit mechanism rather than by conformer selection driven by intrinsic conformational dynamics.

Full text of DOI:10.1021/ja1096539

ARTICLE
pubs.acs.org/JACS
Millisecond Dynamics in Glutaredoxin during Catalytic Turnover Is
Dependent on Substrate Binding and Absent in the Resting States
Kristine Steen Jensen,† Jakob R. Winther, and Kaare Teilum*
Department of Biology, University of Copenhagen, Ole Maaløes Vej 5, 2200 Copenhagen N, Denmark
S Supporting Information
b
ABSTRACT: Conformational dynamics is important for enzy-
me function. Which motions of enzymes determine catalytic
efficiency and whether the same motions are important for all
enzymes, however, are not well understood. Here we address
conformational dynamics in glutaredoxin during catalytic turn-
over with a combination of NMR magnetization transfer, R2
relaxation dispersion, and ligand titration experiments. Glutar-
edoxins catalyze a glutathione exchange reaction, forming a
stable glutathinoylated enzyme intermediate. The equilibrium between the reduced state and the glutathionylated state was
biochemically tuned to exchange on the millisecond time scale. The conformational changes of the protein backbone during catalysis
were followed by 15N nuclear spin relaxation dispersion experiments. A conformational transition that is well described by a two-
state process with an exchange rate corresponding to the glutathione exchange rate was observed for 23 residues. Binding of reduced
glutathione resulted in competitive inhibition of the reduced enzyme having kinetics similar to that of the reaction. This observation
couples the motions observed during catalysis directly to substrate binding. Backbone motions on the time scale of catalytic turnover
were not observed for the enzyme in the resting states, implying that alternative conformers do not accumulate to significant
concentrations. These results infer that the turnover rate in glutaredoxin is governed by formation of a productive enzyme-
substrate encounter complex, and that catalysis proceeds by an induced fit mechanism rather than by conformer selection driven by
intrinsic conformational dynamics.
’ INTRODUCTION
product release as exemplified by ribonuclease A.6 Investigations
of the effect of ligand binding on conformational fluctuations in
dihydrofolate reductase have revealed a dependency on ligand
type.7 Fluctuations in the substrate loop are present only when
substrates are bound to the enzyme but not when cofactors or
products are bound. The observed motions do not correlate with
substrate or cofactor association or dissociation from the
enzyme.8 These findings indicate that microsecond-millisecond
dynamic motions in some enzymes may function by directional
steering of catalysis through conformer selection.7
Determining microsecond-millisecond motions during cata-
lytic turnover results in a more complete description of the
enzymatic energy landscape. NMR relaxation dispersion techni-
ques offer unique possibilities to study exchange processes on the
microsecond-millisecond time scale of proteins in solution at
atomic resolution.9,10 This requires that the exchange rate
between magnetically different states is within the time-window
of the experiments. Minor populated states exchanging with
higher populated states are then probable. For many enzymes
this is not achievable since the equilibrium is shifted far toward
product formation and the reactions, therefore, in practical terms
are irreversible. The dynamics of two types of enzymes, prolyl
cis-trans isomerases and adenylate kinases, however, have been
The link between structure and function of enzymes has been
investigated for decades, and significant insight has been gained
in the coupling of static aspects to enzyme function. The
connection between enzyme function and dynamics, by contrast,
is only partially resolved. Proteins undergo fluctuations on
different time scales, ranging from femtoseconds to picoseconds
for vibrations, and up to hours for rearrangements of subunits
and for folding of some proteins.1 Particular attention has been
paid to microsecond-millisecond conformational motions of
enzymes since macroscopic turnover occurs on a similar time
scale. This similarity suggests that these motions are the rate-
limiting factor of enzyme catalysis.2-4
In purely thermodynamic terms, enzymes catalyze reactions
by decreasing the activation energy for formation of the transi-
tion state. The molecular mechanism underlying this ability is not
fully understood, however. The lifetime of a typical enzyme
transition state is gauged from ab initio transition-path sampling
calculations on the order of femtoseconds.5 Given this short
lifetime and that the transition state should be regarded as highly
dynamic, the much slower microsecond-millisecond conforma-
tional motions cannot directly influence the chemical reaction
coordinate. Motions on these slower time scales can affect the
reaction coordinate in other ways.5 Conformational changes can
facilitate access of the substrate to the catalytic site, control the
correct conformation of the catalytic residues, and promote
Received: October 27, 2010
Published: February 16, 2011
r
2011 American Chemical Society
3034
dx.doi.org/10.1021/ja1096539 J. Am. Chem. Soc. 2011, 133, 3034–3042
|

Journal of the American Chemical Society
ARTICLE
complex of the reduced enzyme with GSH is formed. These
observations indicate that the microsecond-millisecond mo-
tions in hGRX4CS result from substrate binding and not from
crossing of the transition state energy barrier of product forma-
tion. Instead substrate binding induces a conformational change.
Figure 1. Catalytic cycle of glutaredoxin. The reversible reaction of
protein cysteinyl glutathionylation is catalyzed by glutaredoxin (hGRX).
The reaction takes place through two successive thiol-disulfide ex-
change reactions with a stable glutathionylated enzyme intermediate
(hGRXSSG). The shaded box encloses the part of the reaction cycle
studied in the present work, in which glutaredoxin catalyzes the
glutathione interchange.
’ EXPERIMENTAL SECTION
Expression and Purification. A plasmid with cDNA encoding
hGRX4CS in pET24a(þ) (synthesized by GenScript) was used for
expression in E. coli BL21(DE3) Rosetta. All cell growth was at 37 °C.
For production of unlabeled hGRX4CS, cultures were grown in LB
medium and protein expression was induced with IPTG (0.5 mM) at
OD600 = 0.8. Cultures were left for 5 h before harvest. For uniformly 15
N
labeled and 13C,15N labeled hGRX4CS, cultures were first grown in LB
medium to OD600 = 0.8. Then cells were harvested and resuspended in
M9 minimal medium containing 1 g/L 15NH4Cl and 4 g/L glucose
([U-13C6]-glucose) as the sole nitrogen and carbon sources and grown
for 1 h before induction with IPTG. Cells were harvested 4-5 h after
induction. hGRX4CS was purified essentially as described in ref 21 with
the addition of a final RP-HPLC step. For this step, hGRX4CS was
reduced by incubation with DTT (10 mM) for 1 h at room temperature.
TFA was added to the reduced sample to a concentration of 0.5%, and
the sample was eluted from a Zorbax C18 semipreparative column
by a gradient from 30% acetonitrile in 0.1% TFA to 70% acetonitrile in
0.05% TFA. Fractions containing hGRX4CS were flash frozen in liquid
nitrogen and lyophilized. The identity and degree of labeling of
hGRX4CS was confirmed by MALDI-MS. Concentrations of hGRX4CS
were determined using a calculated extinction coefficient at 280 nm of
investigated during catalysis.2,11-15 The results revealed a corre-
lation between catalytic turnover and conformational fluctua-
tions. The frequencies of movements during catalysis occurred
also in the resting enzymes, and both types of enzymes followed
conformational selection where on-pathway conformers were
sampled in the substrate-free state and in the product-bound
state.10,11
The rate of ligand binding may in general also be limited by
conformational changes induced by bound ligand (induced fit) or
by diffusion as inferred by the flexible protein recognition
model.16 The induced fit mechanism of substrate binding lacks
a kinetic description of the rate-limiting step.17 Nevertheless,
discrimination of induced fit from conformer selection requires
kinetic data since the observation of structural change between
ligand-free and ligand-bound enzyme does not tell if the con-
formational change occurred before or after binding of ligand.
The two mechanisms may also be combined where conformer
selection is followed by a ligand -induced structural change.16,18,19
We have investigated motions on the microsecond-millise-
cond time scale of human glutaredoxin 1 (hGRX) during
catalysis and in the resting states. hGRX is an oxidoreductase
localized to the cytosol. It has five cysteine residues of which two
are present in the active site, in a CPYC motif. Other glutar-
edoxins that have a CPYX motif exist.20 The N-proximal cysteine
(Cys23) is the catalytic residue, forming a disulfide bond to
glutathione during the reaction. hGRX4CS is a variant of the
enzyme in which the four noncatalytic cysteine residues are
replaced by serine. Potential glutathionylation of other sites in
the enzyme is thereby prevented. This variant has the same
efficiency as the wild type enzyme in catalysis of deglutathionyla-
tion of glutathionyl mixed disulfides.21 The active site cysteine
has a very low pKa (3.5) ensuring nearly complete deprotonation
of Cys23 at pH 7.22 Because the concentration of the Cys23
anion is high, Cys23 becomes a good leaving group, optimized
for thiol-disulfide exchange.23 In the reversible reaction of
protein glutathionylation, a stable glutathionylated enzyme inter-
mediate is formed (Figure 1). This allows studies of the reaction
with glutathione as the sole substrate. Combined these properties
provide a unique opportunity for studying conformational
dynamics during catalytic turnover in an enzyme under equilib-
rium conditions. The ratio of glutathionylated to reduced
hGRX4CS is tunable by changing the ratio of reduced glutathione
(GSH) to oxidized glutathione (GSSG). The exchange rate
between reduced and glutathionylated hGRX4CS is tunable by
varying the total glutathione concentrations.
2980 M-1cm-1
.
Glutathione Concentration Determinations. Samples of 50
μL were mixed with 50 μL of 1 M HCl and analyzed by RP-HPLC on a
C18 RP-column (Pep-mass) using 0.05% TFA and a linear gradient from
0 to 50% acetonitrile. Absorption was detected at 248 nm. Standard curves
for reduced glutathione (GSH) and oxidized glutathione (GSSG) were
made to link peak areas with concentration. For this purpose the
concentration of a GSH stock was determined in triplicate by reaction
with DTNB (εTNB = 14150 M-1 cm-1), and the GSSG concentration
was determined in triplicate from the absorbance (ε248 = 382 M-1 cm-1).
NMR. All NMR samples were in 10 mM Na-phosphate pH 7.0, 1
mM 4,4-dimethyl-4-silapentane-1-sulfonic acid, and 10% D2O, and all
NMR experiments were performed at 25 °C. NMR relaxation experi-
ments were performed on a Varian Inova 750 MHz spectrometer with a
triple-resonance probe. Assignments and titration experiments were
performed on a Varian Inova 800 MHz spectrometer with a triple-
1
resonance cryoprobe. One-dimensional H spectra were transformed
and analyzed in VnmrJ 2.2D. Multidimensional spectra were processed
in nmrPipe24 and analyzed with nmrPipe and CcpNmr Analysis. Peak
intensities in 1H,15N-HSQC spectra were determined by summing the
intensities in a 3 ꢀ 3 points window centered at the peak maximum.
Assignment of hGRX4CS. Backbone resonances of reduced
hGRX4CS were assigned using 1H-15N HSQC, HNCA, and HNCACB
experiments and the published assignment of reduced wild type
hGRX.25 The sample was 1 mM hGRX4CS with 5 mM DTT. The
assignment of backbone resonances of glutathionylated hGRX4CS could
be transferred from published assignments21 using an HNCA. Glu-
tathionylated glutaredoxin was prepared by incubation of reduced
glutaredoxin with an excess of GSSG (100 mM) at room temperature
for 2 h. Excess glutathione was removed by dialysis (MWCO 3500). The
protein concentration in the final sample was 1 mM.
GSH binding to reduced hGRX4CS was measured by titration of
13C,15N- hGRX4CS with GSH in the presence of 5 mM DTT. GSH was
added directly to the NMR tube from a concentrated stock, and the
sample was purged with argon after each addition. For each
We show that hGRX4CS does not exhibit microsecond-
millisecond conformational changes in the resting states. Mo-
tions are detectable during the reaction and when an off-pathway
3035
dx.doi.org/10.1021/ja1096539 |J. Am. Chem. Soc. 2011, 133, 3034–3042

Journal of the American Chemical Society
ARTICLE
concentration of GSH a 15N-HSQC was recorded. The GSH and
glutaredoxin concentrations after each titration were calculated, ac-
counting for the volume increase. The change in chemical shift was
calculated as the weighted sum of the 1H and 15N chemical shifts relative
glutaredoxin:
hGRX
total
C
K1
½hGRXSHꢁ ¼
ð8Þ
1 þ þ K2½GSHꢁ
to the chemical shifts of free reduced hGRX4CS Δδ = ((δ1H[GSH]
-
Q
δ1H[GSH]=0)2 þ (0.15(δ15N[GSH] - δ15N[GSH]=0))2)1/2. Data were
fitted to eq 1:
Inserting eq 8 into eq 4 and using k-1 = kl/Kl yields
ꢀ
ꢁ
1 þ Q
k0ex ¼ k1chtoGtaRlX
ð9Þ
Q þ K1 þ K2Q½GSHꢁ
Δδmax½GSHꢁ
Δδ ¼
ð1Þ
K2-1 þ ½GSHꢁ
Finally, eq 9 may be inserted into eq 3 to give the following relation
between the measured rates in the magnetization transfer experiment
(λ1 and λ2) and the concentrations of hGRX and glutathione:
Δδmax is the change in chemical shift upon saturation and K2-1 is the
dissociation constant for GSH.
ꢀ
ꢁ
1 þ Q
λ1 þ λ2 ¼ k1cGtortxal
þ R1glutathione
ð10Þ
Magnetization Transfer Analysis of hGRX Catalysis. The
turnover of GSSG to GSH was followed in experiments with hGRX4CS
concentrations of 15, 25, 50, and 100 μM. [GSSG] was kept at 1 mM
while [GSH] was varied (5, 10, 20, and 30 mM). Experiments with 10
mM GSSG and 50 mM GSH were conducted to verify the proposed
model. The exact concentrations of GSSG and GSH in the samples after
equilibration with hGRX4CS, and both before and after the NMR
experiment, were determined as described above. The magnetization
of Hβ1 in GSSG was selectively inverted relative to the Hβ1 of GSH, and
the change in longitudinal magnetization of GSSG Hβ1 was measured in
a series of one-dimensional 1H spectra. The pulse sequence by Robinson
et al.26 with a watergate added was used with relaxation delays (τ) from
0.01 ms to 2 s. The phase cycling was chosen so peak areas go toward
zero at complete relaxation. The peak area, I, of the GSSG specific peak
as function of the relaxation delay was fitted to a biexponential function:
Q þ K1 þ K2Q½GSHꢁ
K1 and k1 were obtained by fitting eq 10 to the experimental data and
setting K2 to the value obtained in the GSH binding experiment.
15N CPMG Relaxation Dispersion. In the 15N Carr-Purcell-
Meiboom-Gill (CPMG) relaxation dispersion experiments the ex-
change rate is determined by the glutathione concentrations:
kex ¼ k1½GSSGꢁ þ k-1½GSHꢁ
ð11Þ
The [GSH]/[GSSG] ratio will control [hGRXSH]/[hGRXSSG] accord-
ing to eq 6.
Three samples were studied: (i) hGRX4CS actively catalyzing glu-
tathione transfer was prepared by dialysis of a 1 mM 15N-hGRX4CS
against a buffer containing 140 μM GSSG and 320 μM GSH. (ii)
Reduced hGRX4CS was prepared by mixing 1 mM 15N-hGRX4CS with 5
mM DTT. (iii) Glutathionylated hGRX4CS was prepared by incubation
of 1 mM reduced 15N-hGRX4CS with 100 mM GSSG and subsequently
removing excess glutathione by RP-HPLC. 15N CPMG relaxation-
compensated experiments27 were performed using the constant-time
approach28 with a relaxation delay of 60 ms. Relaxation dispersions were
sampled using 12 CPMG field strengths (νCPMG) ranging from 50 to
1000 Hz. Data were fitted both to a constant and to the expression for a
two-state process in slow exchange (eq 12).
IðτÞ ¼ A e-λ τ þ B e-λ τ
ð2Þ
1
2
For analyzing the magnetization transfer data, both the reaction
between GSSG and hGRXSH and GSH and hGRXSH were considered as
shown in Schemes 1 and 2:
k1
hGRXSH þ GSSG a hGRXSSG þ GSH ðScheme 1Þ
k-1
k2
hGRXSH þ GSH a hGRXSH GSH
ðScheme 2Þ
3
R2,eff ¼ R2 þ ka½1 - sincðΔω=4νCPMGÞꢁ
ð12Þ
k-2
where R2 is the transverse relaxation in the absence of exchange, ka is the
apparent pseudo-first-order rate constant for the forward (or reverse)
reaction obtained by measuring R2,eff of the reduced (or oxidized) state,
and Δω is the chemical shift difference between the reduced and
oxidized states.
The sum of the eigenvalues (λ1 þ λ2) in eq 2 is equal to the sum of
the exchange rate between reduced and oxidized glutathione (k0ex) and
the average longitudinal relaxation rate for all glutathione species
Rlglutathione
:
λ1 þ λ2 ¼ k0ex þ Rlglutathione
ð3Þ
The F-test was used to determine whether each residue showed
significant dispersions with a significance criterion of p < 0.05, Rex = ka(1
- sinc(Δω/4νCPMG)) > 2 s-1 and ka > χ2/2.
where k0ex is given by
k0ex ¼ k1½hGRXSHꢁ þ k-1½hGRXSSG
ꢁ
ð4Þ
R1F Relaxation Dispersion. A sample of 1 mM 15N-hGRX4CS with
1.7 mM GSH and 5 mM DTT was prepared. Relaxation series with
six time delays (0, 0.02, 0.04, 0.06, 0.08, and 0.1 s) were recorded at
spin-lock field strengths from 406 to 1212 Hz and with offsets ranging
from -1000 to 10 000 Hz. The spin-lock field strengths were calibrated
as described in ref 9. Each relaxation series was fitted to a monoexpo-
nential decay to obtain RlF. The RlF rates were fitted to the expression
for a two-state process in fast exchange (eq 13):
The individual concentrations of hGRXSH and hGRXSSG in the
reaction mixture are too low for accurate determination. Therefore k0ex
should be expressed as a function of the total enzyme concentration:
hGRX
c
¼ ½hGRXSHꢁ þ ½hGRXSSGꢁ þ ½hGRXSH GSHꢁ
ð5Þ
total
3
Rearranging the equilibrium constants K1 and K2 corresponding to
Schemes 1 and 2, respectively, gives
R1F ¼ R1 cos2 θ þ ðR2 þ RexÞsin2 θ
ð13Þ
K1
½hGRXSSGꢁ ¼ ½hGRXSH
ꢁ
ð6Þ
ð7Þ
2
where θ is the tilt angle, Rex = Φexkex/(kex þ ωeff2), kex is the exchange
rate between the two states, ωeff is the effective field of the spin-lock, Φex
= Δω2pA(1 - pA), Δω is the chemical shift difference between the
exchanging states, and pA is the relative population of one of the states.
The F-test (p < 0.01) was used to determine if the Rex contribution was
significant by comparing fits to eq 13 where Rex was either optimized or
fixed at Rex = 0.
Q
½hGRXSH GSHꢁ ¼ K2½GSHꢁ½hGRXSH
ꢁ
3
where Q = [GSH]/[GSSG]. Substituting these two expressions into eq 5
yields the following expression for the concentration of reduced
3036
dx.doi.org/10.1021/ja1096539 |J. Am. Chem. Soc. 2011, 133, 3034–3042

Journal of the American Chemical Society
ARTICLE
Isothermal Titration Calorimetry. Reduced hGRX4CS was
titrated with GSSG at 25 °C on a MCS isothermal titration calorimeter
(MicroCal LLC, Northampton, MA). Prior to the titration, hGRX4CS
was reduced by incubation with 100 mM DTT, 10 mM Na-phosphate,
pH 7.0, for 2 h at room temperature followed by overnight incubation at
4 °C. DTT was removed by gel filtration on a PD-10 Sephadex-G25
column using argon-saturated 10 mM Na-phosphate, pH 7.0, as eluent.
The fraction containing reduced glutaredoxin was concentrated by use
of spin filters (MWCO 3000) to 165 μM. The GSSG stock of 1.60 mM
was prepared in 10 mM Na-phosphate, pH 7.0. Glutaredoxin and GSSG
stocks were thoroughly purged with argon and degassed by vacuum
stirring for 20 min before hGRX4CS was loaded into the ITC cell and
GSSG was loaded in the syringe. Data were analyzed using an exchange
model describing the reaction in Scheme 1 assuming that the measured
change in enthalpy (ΔH) equals the change in standard enthalpy
(ΔH°).29
’ RESULTS
Characterization of the Glutaredoxin-Glutathione Equi-
librium. To choose conditions that allow delineation of con-
formational fluctuations in hGRX4CS during catalysis, the
reaction needs to be characterized. This was accomplished using
NMR magnetization transfer30 at chemical equilibrium to de-
termine the rate constants of the reaction between glutathione
and glutaredoxin (Scheme 1). The relaxation rates comprise
contributions from longitudinal relaxation and chemical ex-
change (Figure 2A and Scheme 1). A specific signal from one
of the exchanging species is required, implying that the reaction
has to be in slow exchange on the chemical shift time scale. The
reaction was followed by measuring the change in magnetization
of the Hβ resonance of the cysteinyl residue of glutathione, which
splits upon formation of the disulfide bond present in GSSG.31
An experiment series with varied enzyme concentrations and
different GSH to GSSG ratios (Q) was conducted (Figure 2).
High concentrations of GSH resulted in slower rates than
expected from a single equilibrium between glutaredoxin and
glutathione (Scheme 1) and thus made the apparent forward rate
constant dependent on the GSH concentration. The simplest
model to explain the observed behavior is that GSH may also
bind to reduced hGRX4CS and thereby decrease the effective
enzyme concentration (Scheme 2).
Figure 2. Determination of rate constants by magnetization transfer.
(A) The rate constants are determined by observation of the glutar-
edoxin-catalyzed magnetization transfer between GSSG and GSH.
Stars indicate the glutathione that originates in the oxidized state and
emphasize the process of exchange between GSSG and GSH, which is
catalyzed by hGRX. (B) Peak area of cysteinyl Hβ specific for GSSG
(3.31 ppm) plotted as a function of the relaxation delay (τ) for Q = 20.
Solid lines are fits to eq 2. The experiment was performed with 1 mM
GSSG and 20 mM GSH at 25 °C, pH 7.0. (C) Data are plotted according
to a model describing the glutaredoxin-glutathione equilibrium as
the coupling of Schemes 1 and 2. X-axis values are calculated using
the fitted value of Kl and the value of K2 obtained from the GSH titration.
Error bars show standard errors from fits of eq 2 to the data. The straight
line is drawn using eq 10 and the fitted values for kl ((1.24 ( 0.03) ꢀ
106 M-1 s-1) and R1,glutathione (4.2 ( 0.3 s-1).
To test this model, reduced hGRX4CS was titrated with GSH,
and 15N-1H amide chemical shifts changes were followed.
Resonances experiencing a weighted chemical shift difference
of at least 0.1 ppm were included in a global fit of the dissociation
constant (Figure 3). All titrating resonances report on the same
event and result in a dissociation constant of K2-1 = 15.4 ( 0.3
mM. A model containing the reactions of both Schemes 1 and 2 is
thus necessary to describe the system (eq 10). Increasing the
total concentration of glutathione ten times while keeping the
ratio of GSH to GSSG near five did not result in deviation from
the model. The equilibrium constants of the two reactions, Kl and
K2, are strongly correlated in eq 10 and cannot be determined
independently from fits of this equation to the experimental data.
K2 was fixed, therefore, to the value from the titration of reduced
hGRX4CS with GSH. Values of Kl, kl, and the average longitudinal
relaxation rate of glutathione Rlglutathione were optimized by
fitting eq 10 to the kinetic data. The fit resulted in a forward
rate constant k1 of (1.24 ( 0.03) ꢀ 106 M-1 s-1, an equilibrium
constant K1 of 29.1 ( 1.4, and a calculated backward rate
constant k-1 of (4.3 ( 0.7) ꢀ 104 M-1 s-1. K1 was also
Figure 3. NMR titration of reduced hGRX4CS with GSH. The experi-
ment was conducted at 25 °C, Na-phosphate pH 7.0 with a 1 mM
13C,15N-hGRX4CS sample. Changes in chemical shift were calculated as
a weighted sum of the change in both direct and indirect dimension; Δδ =
((δ1H[GSH] - δ1H[GSH]=0)2 þ (0.15(δ15N[GSH] - δ15N[GSH]=0))2)1/2
.
The 23 resonances displaying a chemical shift difference larger than 0.1 ppm
were included in a global fit to eq 1, which is represented by the solid lines.
3037
dx.doi.org/10.1021/ja1096539 |J. Am. Chem. Soc. 2011, 133, 3034–3042

Journal of the American Chemical Society
ARTICLE
determined by isothermal titration calorimetry yielding an
equilibrium constant of 35.1 ( 0.6, which is in good agreement
with K1 obtained from the NMR data. The analysis shows that
three different states of hGRX4CS are populated at equilibrium,
and that the enzyme cannot be saturated with either of the
substrates at concentrations as high as 50 mM GSH and 10 mM
GSSG (Figure 4).
Structural Differences between Observed Glutaredoxin
States. The three observed states of glutaredoxin in equilibrium
with glutathione are (i) the reduced enzyme, (ii) the glutathio-
nylated enzyme, in which glutathione is covalently attached to
hGRX4CS through a disulfide bond to Cys23, and (iii) the
complex between reduced hGRX4CS and GSH. The structures
of reduced wild type hGRX and glutathionylated hGRX4CS are
very similar with an overall rmsd between the mean structures of
0.96 Å.21,25 Calculation of the mean displacement of each CR atom
relative to all other CR atoms between the two structures revealed
that the largest differences are in the loop regions and the
C-terminal end (Figure 5A,B). Comparing the 15N chemical shifts
of reduced hGRX4CS with the 15N chemical shifts of glutathiony-
lated hGRX4CS shows significant differences (Figure 5C,D). The
largest were for Lys20, Cys23, Arg28, Ala29, Ala67, Arg68, Thr69,
Val70, Ile80, and Ser83. Panels A and C in Figure 5 show that
chemical shift differences around residues Cys23 and Thr69
correspond to the largest structural changes of up to 1.4 Å while
the chemical shift change for Ser83 is associated with a mean
backbone displacement of only 0.5 Å. Chemical shift data for
Asn52 are not available, as the resonance was missing from the
spectra. From the titration of reduced hGRX4CS with GSH,
structural information about the GSH binding site was obtained.
The largest differences were for Lys20, Thr22, and Val70. The
change in chemical shift of Cys23 is, not surprisingly, much less
than when glutathione is covalently bound. The changes in the
chemical shifts of the other residues are also less pronounced. Still,
the pattern shows that GSH binds to reduced hGRX4CS in the
binding site of covalently attached glutathione. The smaller
chemical shift changes for the binding of GSH also indicate that
the mere presence of the ligand is insufficient to account for all
changes in chemical shifts observed between free and glutathio-
Figure 4. Schematic overview of different states of glutaredoxin
(hGRX) during the reaction with glutathione. Three different states
are experimentally observed corresponding to the reduced enzyme
(hGRXSH), the glutathionylated enzyme (hGRXSSG), and the GSH
bound reduced enzyme (hGRXSH GSH). The reaction between
3
hGRXSH and oxidized glutathione (GSSG) must proceed though the
putative complexes [hGRXSH GSSG] and [hGRXSSG GSH] although
3
3
these are not observed experimentally.
nylated hGRX4CS
.
Figure 5. 15N chemical shift differences between glutaredoxin states. (A) Mean CR displacement calculated as the mean of residue-by-residue CR-CR
position difference between the lowest energy structure of glutathionylated hGRX4CS and of reduced hGRX. (B) Alignment of the 20 NMR structures of
lowest energy of glutathionylated hGRX4CS (cyan) (PDB id: 1B4Q) with the 20 lowest energy structures of reduced wild type hGRX (dark blue) (PDB id:
1JHB), with a root-mean-square deviation of 0.96 Å from alignment with Pymol (DeLano Scientific). (C) Chemical shift differences between reduced
glutaredoxin (1 mM 15N-hGRX4CS, 5 mM DTT) and glutathionylated glutaredoxin (15N-hGRX4CS) and between reduced glutaredoxin (15N-hGRX4CS, 5
mM DTT) and the complex ofreduced glutaredoxin with GSH (13C,15N-hGRX4CS, 5 mM DTT, 128 mM GSH) are shown. All spectra were recorded at25
°C, pH 7.0. Differences are observable in the same areas although largest when glutathione is covalently attached to Cys23 through a disulfide bond. The
secondary structure of glutaredoxin is displayed on top. (D) Cartoon representation of glutathionylated glutaredoxin with glutathione displayed in sticks
(PDB id: 1B4Q). The structure is colored according to the chemical shift difference between reduced hGRX4CS and glutathionylated hGRX4CS with a color
gradient running from dark blue over white to dark red with 29 colors with dark blue corresponding to less than 0.005 ppm difference and dark red
corresponding to more than 1 ppm difference. Black denotes proline residues and residues not assigned in at least one of the states.
3038
dx.doi.org/10.1021/ja1096539 |J. Am. Chem. Soc. 2011, 133, 3034–3042

Journal of the American Chemical Society
ARTICLE
Figure 6. 15N,1H-HSQC glutathione reaction-sample used in the
CPMG experiment. Peaks from the glutathionylated state are red and
from the reduced state blue. Peaks with spectral overlap and unassigned
peaks are black. No relaxation dispersion enhancement was found for
any of the unassigned peaks. The experiments were recorded at 25 °C,
pH 7.0, on 1 mM 15N-hGRX4CS. The reaction sample contained 22 μM
GSSG and 560 μM GSH.
Microsecond-Millisecond Conformational Fluctuations
in Glutaredoxin. Backbone motions in glutaredoxin during
catalysis and in the two resting states were investigated by CPMG
15N spin relaxation dispersion experiments. To measure dy-
namics in hGRX4CS related to catalysis, the concentrations of
GSH and GSSG were adjusted to reach slow exchange between
the reduced and glutathionylated enzyme on the NMR time
scale. Therefore, cross-peaks originating from backbone amides
in both exchanging states are visible in the spectra (Figure 6), and
the relaxation dispersions become independent of field
strength.32 In the resting states, no relaxation dispersions are
observed, demonstrating that the backbone of the enzyme dis-
plays no significant dynamics on the millisecond time scale
(Figure 7). The reaction sample was set up with 1 mM hGRX4CS
,
Figure 7. Microsecond-millisecond conformational motions in glutar-
edoxin. (A) Chemical shift correlation of Δω from pairwise fits of R2
dispersion data out of the glutathionylated state and out of the reduced
state with the observed chemical shift difference between the two resting
states (Δδ). The linear fit with 95% confidence intervals is shown. (B)
R2 dispersion of catalytic active Cys23 during reaction (filled circles) and
in the two resting states: reduced (open triangles), glutathionylated
(filled triangles). The red curve is a global fit of eq 12 to the reaction data.
Data from the resting states are fitted to a constant. Errors are estimated
from the signal-to-noise level in the spectra. (C) R2 dispersion of Ala29
during reaction and in the two resting states. Symbols and curves as for
B. The experiments were recorded at 25 °C, pH 7.0, on 1 mM 15N-
hGRX4CS. The reaction sample contained 22 μM GSSG and 560 μM
GSH. The reduced sample contained 5 mM DTT.
320 μM free GSH, and 140 μM free GSSG by dialysis. These
concentrations correspond to a population of about ten percent
reduced hGRX4CS and kex of 190 s-1 as gauged from eqs 6 and
11. In the heteronuclear single quantum coherence (HSQC)
spectrum of the sample, cross-peaks from a number of residues
were identified as originating from either the glutathionylated
state or the reduced state by comparison with spectra of the free
forms (Figure 6). The chemical shift differences (Δω) obtained
by fitting a two-state exchange model (eq 12) to the relaxation
dispersion data correlate with the chemical shift difference
between the two resting states (Δδ) (Figure 7A and Table S1,
Supporting Information) and reveal that no intermediate states
are needed to explain the experimental data. At the chosen
suppressed even at the highest CPMG pulsing frequency. This
behavior results mainly from the large chemical shift changes of
these residues. That these residues located in the active site may
experience additional dynamics cannot be ruled out. Asp59 is not
included, as it is in intermediate exchange, and the parameters are
not well-defined by the experiment.
conditions the hGRXSH GSH off-path state was populated to
3
less than one percent, and therefore not detectable in the
experiment. Significant dispersions in 15N-backbone R2 relaxa-
tion rates were found for 23 residues of which nine have the NH
atoms within 6 Å of bound glutathione. The direct chemical
effect from electrons in the substrate must be large on these
residues (Figures 7and 8; Figure S1 and Table S2, Supporting
Information). The largest dispersions were seen for Val70 and
Cys23, as expected from the chemical shift differences between
hGRXSH and hGRXSSG (Figures 5 and 7). Dispersions were also
detected for Tyr25, Ser26, and Ser83, but the analysis is unreli-
able since the effect of the exchange process is not fully
A global fit of kaox and kared (eq 12) to the dispersion data gives
kex of 51.2 ( 1.2 s-1, and a population of hGRXSH of 47% (Table
S2). This result corresponds to [GSH] = 560 μM and [GSSG] =
22 μM and shows that an undesired shift in the equilibrium had
occurred during sample preparation. At the low glutathione
concentrations used in the experiment, accurate determination
of the concentrations directly in the sample are not possible.
3039
dx.doi.org/10.1021/ja1096539 |J. Am. Chem. Soc. 2011, 133, 3034–3042

Journal of the American Chemical Society
ARTICLE
Figure 9. Conformational fluctuations of glutaredoxin upon binding of
GSH. Relaxation dispersion profile of R2,eff = R2 þ Rex (extracted from
eq 13) for Val70 as function of the effective field strength. The solid line
represents a fit of the relaxation dispersion data to eq 13 with kex as a
global parameter for the three residues that show exchange (Lys20,
Thr22, and Val70). φex were fixed to the calculated values based on the
GSH titration data. Standard errors on R2,eff are propagated from eq 13.
The experiment was performed on a sample of 1 mM 15N-hGRX4CS with
1.7 mM GSH, 5 mM DTT, pH 7.0, at 25 °C.
Dynamics in the backbone associated with binding of GSH to
the reduced enzyme were investigated by off-resonance rotating-
frame (RlF), relaxation dispersion measurements. Using the
determined value of K2, a GSH concentration of 1.7 mM was
chosen to populate the minor state (hGRX4CS GSH) to 10%.
3
Figure 8. Residues with enhanced relaxation dispersion rates. Residues
showing enhanced amide R2 dispersion during reaction are displayed
with backbone CR atoms in the space-filling model. Residues with NH
within 6 Å from atoms in bound glutathione are magenta, residues with
NH more than 6 Å away but less than 8 Å are pink, and the remainder are
red. Red, pink, and magenta residues are included in the global fit of kex.
Tyr25, Ser26, Asp59, and Ser83 are colored gray, as they are not included
in the global fit. The displayed structure is the lowest energy structure
of glutathionylated hGRX4CS (PDB id: 1B4Q) with glutathione in balls
and sticks.
Small, yet significant, relaxation dispersions were observable for
the three amides with the largest chemical shift differences
(Figure 9). As expected from the GSH-induced chemical shift
changes on hGRXSH (Figure 5C), no significant relaxation
dispersion was detected for Cys23. A global fit of kex (eq 13)
to the dispersion curves of Lys20, Thr22, and Val70 yields kex of
10510 ( 720 s-1. Using this value of kex in combination with K2
gives the rate constants k2 of (0.61 ( 0.04) ꢀ 106 M-1 s-1 and
k-2 of (9.5 ( 0.6) ꢀ 103 s-1. The on-rate constant, k2, is within
a factor of 2 similar to the forward rate constant k1, for GSSG
turnover. Binding of GSH in the active site thus results in
dynamics in hGRX4CS similar to that observed during catalysis.
[GSH] was estimated to 300-650 μM by HPLC whereas
[GSSG] was below 100 μM, which is consistent with the
glutathione concentrations calculated from the kinetics. The
shift in the equilibrium does not influence the conclusions drawn
from the data. Precautions were taken to avoid oxygen in the
sample, as oxidation may result in drifting of the glutathione
concentrations. No intensity differences, however, were observed
in HSQC spectra before and after the NMR experiment,
demonstrating that the glutathione concentrations were constant
during the experiment.
Binding Dynamics of Glutaredoxin. Because motions were
not detected in the resting states of the enzyme, the observed
motions could be linked to transition state barrier crossing or to
binding of substrates. The observation that GSH binds to the
reduced enzyme in the binding site of covalently bound
glutathione permitted discrimination between these two op-
tions. That residues showing exchange during catalysis are the
same as those that change their chemical shifts when hGRXSH
becomes glutathionylated and when GSH is added to hGRXSH
suggest that only one-half of GSSG makes significant interac-
tions with the enzyme. Binding of GSH and GSSG should
occur, therefore, with comparable rates. If binding is limiting,
we will consequently expect to observe comparable rates for
GSH binding and turnover. If on the other hand transition state
barrier crossing is limiting, GSH binding should be faster than
turnover.
’ DISCUSSION
Conformational Fluctuations Are Linked to Binding. The
analysis of slow microsecond-millisecond backbone conforma-
tional fluctuations in hGRX4CS during catalysis of glutathione
interchange shows that kex calculated from a global fit of the R2
dispersion curves agrees with the catalytic turnover rate. The
processes detected by the enzyme and by the substrate must,
therefore, be limited by the same event. The correlation of the
chemical shift difference between the two exchanging states and
that between reduced and glutathionylated hGRX4CS reveals that
the two exchanging states correspond to the reduced and
glutathionylated states of the enzyme. This is characteristic for
a two-state process. These data alone do not reveal whether the
kinetics of the fluctuations is limited by binding alone or whether
crossing of the catalytic energy barrier to product formation also
contributes.
The finding that GSH binds to reduced hGRX4CS forming
an off-pathway complex allowed testing if the conformational
fluctuations in glutaredoxin are directly induced by binding.
Analysis of microsecond-millisecond fluctuations in reduced
hGRX4CS during binding of GSH revealed motions similar
to those observed for hGRX4CS during catalysis. The rate
3040
dx.doi.org/10.1021/ja1096539 |J. Am. Chem. Soc. 2011, 133, 3034–3042

Journal of the American Chemical Society
ARTICLE
constant for binding of GSH (k2) is highly similar to kl differing
only by a factor of 2. The similar kinetics of the two processes
suggest that the conformational changes in hGRX4CS observed
during catalysis are linked directly to binding of glutathione
and that no intermediates populate to an extent that influences
the rate of product formation. In contrast, the rate of dissocia-
Fluctuations between different conformers of hGRX may
occur on even faster time-scales than probed here. Indeed, the
order parameters for reduced E. coli GRX1 are lower than those
for the oxidized state, implying that the fast (picosecond-
nanosecond) dynamics has a higher amplitude in the reduced
state than in the oxidized state.34 This fast dynamics cannot be
limiting for substrate binding and catalytic turnover, however. A
fast pre-equilibrium between a major state and a binding-
competent minor state would result in lowering the effective
concentration of the active enzyme, implying that the enzyme
had to be more efficient than if there were no pre-equilibrium.
Induced fit and conformational selection are two extreme
models for ligand binding. Some elements of conformer selection
likely occur in binding reactions that appear to follow an induced
fit model (and vise versa) even if they are not observed.1,16,19 Our
results only describe backbone fluctuations. Specific side-chain
conformations may still be selected as found for FKBP12 where
backbone and side-chain microsecond-millisecond fluctuations
are independent in a complex with the transition state analogue
FK506.35 It remains to be investigated whether the latter is also
the case for hGRX4CS, allowing for an element of side-chain
conformer selection in addition to the predominating induced fit
mechanism for substrate binding.
tion of hGRX4CS GSH (k-2 = 9.5 ꢀ 103 s-1) is much faster
3
than the concentration-dependent rate of the bimolecular
reaction of deglutathionylation of hGRX4CS (Scheme 1),
which takes place at a rate of k-1 ꢀ [GSH] (for [GSH] =
320 μM the rate is 14 s-1).
The experiments probe changes in the magnetic environment
although contributions from conformational fluctuations cannot
be distinguished from the sole effect of ligand presence. There-
fore, the observed dispersion on residues close to the ligand may
contain contributions from both. Nine of the twenty-three
residues showing exchange during catalysis have NH atoms
within 6 Å from bound glutathione whose chemical shifts will
be highly dependent on the change in electronic environment
from binding alone (Figure 8). Electrostatic effects may reach
further than 6 Å but will be small and difficult to predict. Ten of
the twenty-three residues (Phe18, Arg28, Ala29, Ile48, Thr49,
Asp59, Gln62, Gly76, Gly80, and Val87) have NH atoms more
than 8 Å away from the ligand. Enhanced dispersions on these
residues likely reflect conformational motions. All together the
present results show that microsecond-millisecond conforma-
tional fluctuations are present in hGRX4CS only during catalysis
of the glutathione interchange and when the reduced enzyme
binds reduced glutathione. These observations infer that the
observed motions, and thus the change in conformation between
the free and bound forms of hGRX4CS, are a direct result of
binding glutathione and that binding occurs through an induced
fit mechanism.
Lack of Intrinsic Microsecond-Millisecond Backbone Dy-
namics in Glutaredoxin. The finding that microsecond-milli-
second conformational fluctuations are not detectable in the
resting states of hGRX4CS distinguishes hGRX4CS from many
other investigated enzymes.10 NMR dispersion data yield neither
information about amplitude nor direction of an observed
fluctuation. This information can be obtained by coupling
chemical shift of interchanging states to structural data. Intrinsic
microsecond-millisecond dynamics in adenylate kinase is asso-
ciated with large loop movements whereas movements in RNase
A and dihydrofolate reductase are of smaller amplitude.8,13,33
Collectively these fluctuations allow for substrates and products
to access the active site. In all three enzymes, product release is
rate-limiting for catalysis. The active site of hGRX4CS is not
protected by a loop, and the absence of resting state backbone
fluctuations reflects the direct access to the active site.
Catalysis by Glutaredoxin. In conclusion, the lack of resting
state conformational fluctuations combined with the finding that
kex obtained from the relaxation dispersion data corresponds to
kex of the reaction suggests that the turnover rate of glutaredoxin
is controlled by formation of a productive encounter complex
between enzyme and substrate, and that the conformational
changes subsequently occur by an induced fit mechanism. The
latter cannot be rate-limiting, as no indication of substrate
saturations were observed in accord with previous kinetic
analysis showing that saturation was impossible.22 The identified
microscopic rate constants are consistent with earlier interpreta-
tions identifying the reaction between glutathionylated glutar-
edoxin and GSH as the rate-limiting step in the catalytic cycle of
hGRX4CS 22
The present data fit a two-state model in which
.
population of an encounter complex is not detectable, reflecting
high commitment to catalysis. Our analysis of hGRX4CS catalysis
of the glutathione interchange reaction illustrates that conforma-
tional dynamics need not control the catalytic mechanism of an
enzyme and that hGRX4CS functions by providing optimal
conditions for thiol-disulfide exchange. A general unifying
model to explain the connection between conformational dy-
namics in enzymes and catalysis will have to account for different
rate-limiting events in different enzymes.1,16,19
’ ASSOCIATED CONTENT
A consequence of missing dynamics in the resting states
comparable to the turnover rate is that hGRX4CS cannot follow
a mechanism where catalysis is controlled by conformer selection
in substrate binding or product release. The latter is supported by
the failure to saturate hGRX with substrate concentrations up to
50 mM GSH and 10 mM GSSG. If the chemical shift differences
associated with fluctuations between the two states are very small
in the absence of ligand, however, they may not be probed by the
CPMG experiment. Likewise, the population of a higher energy
state may also be so low that it is nonprobable by the NMR
experiment. The involvement of such a high energy binding
competent state in hGRX4CS can be excluded, as substrate
saturation was not observed.
S
Supporting Information. Figure S1: R2 relaxation dis-
b
persion plots of residues showing enhanced dispersion during
reaction. Table S1: Fitting parameters for pairwise fit of disper-
sion data. Table S2: Fitting parameters from global fit of the
relaxation dispersion data. This material is available free of charge
via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION
Corresponding Author
kaare.teilum@bio.ku.dk
3041
dx.doi.org/10.1021/ja1096539 |J. Am. Chem. Soc. 2011, 133, 3034–3042

Journal of the American Chemical Society
ARTICLE
Present Addresses
(26) Robinson, G.; Kuchel, P. W.; Chapman, B. E.; Doddrell, D. M.;
Irving, M. G. J. Magn. Reson. 1985, 63, 314–319.
(27) Loria, J. P.; Rance, M.; Palmer, A. G. J. Am. Chem. Soc. 1999,
121, 2331–2332.
(28) Tollinger, M.; Skrynnikov, N. R.; Mulder, F. A.; Forman-Kay,
J. D.; Kay, L. E. J. Am. Chem. Soc. 2001, 123, 11341–11352.
(29) Iversen, R.; Andersen, P. A.; Jensen, K. S.; Winther, J. R.;
Sigurskjold, B. W. Biochemistry 2010, 49, 810–820.
(30) Led, J. J.; Gesmar, H.; Abildgaard, F. Methods Enzymol. 1989,
176, 311–329.
(31) Rabenstein, D. L.; Millis, K. K. Biochim. Biophys. Acta 1995,
1249, 29–36.
(32) Millet, O.; Loria, J. P.; Kroenke, C. D.; Pons, M.; Palmer, A. G.
J. Am. Chem. Soc. 2000, 122, 2867–2877.
(33) Beach, H.; Cole, R.; Gill, M. L.; Loria, J. P. J. Am. Chem. Soc.
†Chemistry Department, Institute for Biophysical Chemistry,
TU M€unchen, Lichtenbergstrasse 4, D-85748 Garching,
Germany.
’ ACKNOWLEDGMENT
This work was supported by the Danish Research Agency for
Nature and Universe grant number 272-06-0251 (K.T.). K.S.J.
received a stipend from Faculty of Science, University of
Copenhagen. Dr. Bent W. Sigurskjold is thanked for an intro-
duction to the isothermal titration calorimetry setup. We thank
Penny von Wettstein-Knowles for critically reviewing the manu-
script of this paper.
2005, 127, 9167–9176.
(34) Kelley, J. J., III; Caputo, T. M.; Eaton, S. F.; Laue, T. M.;
Bushweller, J. H. Biochemistry 1997, 36, 5029–5044.
(35) Brath, U.; Akke, M. J. Mol. Biol. 2009, 387, 233–244.
’ REFERENCES
(1) Teilum, K.; Olsen, J. G.; Kragelund, B. B. Cell. Mol. Life Sci. 2009,
66, 2231–2247.
(2) Wolf-Watz, M.; Thai, V.; Henzler-Wildman, K.; Hadjipavlou, G.;
Eisenmesser, E. Z.; Kern, D. Nat. Struct. Mol. Biol. 2004, 11, 945–949.
(3) Doucet, N.; Watt, E. D.; Loria, J. P. Biochemistry 2009, 48, 7160–
7168.
(4) Kovrigin, E. L.; Loria, J. P. J. Am. Chem. Soc. 2006, 128, 7724–
7725.
(5) Schwartz, S. D.; Schramm, V. L. Nat. Chem. Biol. 2009, 5, 551–
558.
(6) Kovrigin, E. L.; Loria, J. P. Biochemistry 2006, 45, 2636–2647.
(7) Boehr, D. D.; McElheny, D.; Dyson, H. J.; Wright, P. E. Proc.
Natl. Acad. Sci. U.S.A. 2010, 107, 1373–1378.
(8) Boehr, D. D.; McElheny, D.; Dyson, H. J.; Wright, P. E. Science
2006, 313, 1638–1642.
(9) Palmer, A. G., III; Kroenke, C. D.; Loria, J. P. Methods Enzymol.
2001, 339, 204–238.
(10) Kern, D.; Eisenmesser, E. Z.; Wolf-Watz, M. Methods Enzymol.
2005, 394, 507–524.
(11) Eisenmesser, E. Z.; Millet, O.; Labeikovsky, W.; Korzhnev,
D. M.; Wolf-Watz, M.; Bosco, D. A.; Skalicky, J. J.; Kay, L. E.; Kern, D.
Nature 2005, 438, 117–121.
(12) Eisenmesser, E. Z.; Bosco, D. A.; Akke, M.; Kern, D. Science
2002, 295, 1520–1523.
(13) Henzler-Wildman, K. A.; Thai, V.; Lei, M.; Ott, M.; Wolf-Watz,
M.; Fenn, T.; Pozharski, E.; Wilson, M. A.; Petsko, G. A.; Karplus, M.;
Hubner, C. G.; Kern, D. Nature 2007, 450, 838–844.
(14) Henzler-Wildman, K. A.; Lei, M.; Thai, V.; Kerns, S. J.; Karplus,
M.; Kern, D. Nature 2007, 450, 913–916.
(15) Labeikovsky, W.; Eisenmesser, E. Z.; Bosco, D. A.; Kern, D.
J. Mol. Biol. 2007, 367, 1370–1381.
(16) Grunberg, R.; Leckner, J.; Nilges, M. Structure 2004, 12, 2125–
2136.
(17) Koshland, D. E. Proc. Natl. Acad. Sci. U.S.A 1958, 44, 98–104.
(18) Boehr, D. D.; Nussinov, R.; Wright, P. E. Nat. Chem. Biol. 2009,
5, 789–796.
(19) Csermely, P.; Palotai, R.; Nussinov, R. Trends Biochem. Sci.
2010, 35, 539–546.
(20) Lillig, C. H.; Berndt, C.; Holmgren, A. Biochim. Biophys. Acta
2008, 1780, 1304–1317.
(21) Yang, Y.; Jao, S.; Nanduri, S.; Starke, D. W.; Mieyal, J. J.; Qin, J.
Biochemistry 1998, 37, 17145–17156.
(22) Srinivasan, U.; Mieyal, P. A.; Mieyal, J. J. Biochemistry 1997, 36,
3199–3206.
(23) Jensen, K. S.; Hansen, R. E.; Winther, J. R. Antioxid. Redox.
Signal. 2009, 11, 1047–1058.
(24) Delaglio, F.; Grzesiek, S.; Vuister, G. W.; Zhu, G.; Pfeifer, J.;
Bax, A. J. Biomol. NMR 1995, 6, 277–293.
(25) Sun, C.; Holmgren, A.; Bushweller, J. H. Protein Sci. 1997, 6,
383–390.
3042
dx.doi.org/10.1021/ja1096539 |J. Am. Chem. Soc. 2011, 133, 3034–3042