Modulation of a pre-existing conformational equilibrium tunes adenylate kinase activity

Article abstract of DOI:10.1021/ja3032482

Structural plasticity is often required for distinct microscopic steps during enzymatic reaction cycles. Adenylate kinase from Escherichia coli (AK_eco) populates two major conformations in solution; the open (inactive) and closed (active) state, and the overall turnover rate is inversely proportional to the lifetime of the active conformation. Therefore, structural plasticity is intimately coupled to enzymatic turnover in AK_eco. Here, we probe the open to closed conformational equilibrium in the absence of bound substrate with NMR spectroscopy and molecular dynamics simulations. The conformational equilibrium in absence of substrate and, in turn, the turnover number can be modulated with mutational- and osmolyte-driven perturbations. Removal of one hydrogen bond between the ATP and AMP binding subdomains results in a population shift toward the open conformation and a resulting increase of k_cat. Addition of the osmolyte TMAO to AK_eco results in population shift toward the closed conformation and a significant reduction of k_cat. The Michaelis constants (K_M) scale with the change in k_cat, which follows from the influence of the population of the closed conformation for substrate binding affinity. Hence, k_cat and K_M are mutually dependent, and in the case of AK_eco, any perturbation that modulates k_cat is mirrored with a proportional response in K_M. Thus, our results demonstrate that the equilibrium constant of a pre-existing conformational equilibrium directly affects enzymatic catalysis. From an evolutionary perspective, our findings suggest that, for AK_eco, there exists ample flexibility to obtain a specificity constant (k_cat/K_M) that commensurate with the exerted cellular selective pressure.

Full text of DOI:10.1021/ja3032482

Article
pubs.acs.org/JACS
Modulation of a Pre-existing Conformational Equilibrium Tunes
Adenylate Kinase Activity
†
‡
‡
,†
Jo
†
̈
rgen Åden
́
, Abhinav Verma, Alexander Schug, and Magnus Wolf-Watz*
Department of Chemistry, Chemical Biological Center, Umea
Steinbuch Centre for Computing, Karlsruhe Institute of Technology, 76131 Karlsruhe, Germany
S Supporting Information
̊ ̊
University, SE-901 87 Umea, Sweden
‡
*
ABSTRACT: Structural plasticity is often required for distinct
microscopic steps during enzymatic reaction cycles. Adenylate
kinase from Escherichia coli (AK ) populates two major
eco
conformations in solution; the open (inactive) and closed
(
active) state, and the overall turnover rate is inversely
proportional to the lifetime of the active conformation.
Therefore, structural plasticity is intimately coupled to
enzymatic turnover in AK . Here, we probe the open to
eco
closed conformational equilibrium in the absence of bound
substrate with NMR spectroscopy and molecular dynamics
simulations. The conformational equilibrium in absence of substrate and, in turn, the turnover number can be modulated with
mutational- and osmolyte-driven perturbations. Removal of one hydrogen bond between the ATP and AMP binding subdomains
results in a population shift toward the open conformation and a resulting increase of k . Addition of the osmolyte TMAO to
cat
AKeco results in population shift toward the closed conformation and a significant reduction of k . The Michaelis constants (K )
cat
M
scale with the change in k , which follows from the influence of the population of the closed conformation for substrate binding
cat
affinity. Hence, kcat and K are mutually dependent, and in the case of AK , any perturbation that modulates k is mirrored
M
eco
cat
with a proportional response in K . Thus, our results demonstrate that the equilibrium constant of a pre-existing conformational
M
equilibrium directly affects enzymatic catalysis. From an evolutionary perspective, our findings suggest that, for AK , there exists
eco
ample flexibility to obtain a specificity constant (k /K ) that commensurate with the exerted cellular selective pressure.
cat
M
INTRODUCTION
AKeco catalyzes the magnesium-dependent reversible intercon-
version of AMP and ATP into two ADP molecules and harbors
distinct ATP and AMP binding subdomains (ATPlid and
AMPbd). The global stability of the enzyme is governed by the
■
Enzymes are extraordinary biocatalysts that can increase the
rate of chemical reactions several orders of magnitude, making
chemical reactions to occur on a time scale that is compatible
with global biological processes such as cell division. For
instance, the rate enhancement of orotidine 5′-phosphate
decarboxylase is estimated to be 10 -fold. Many of the
microscopic steps during enzymatic reaction cycles are
dependent on structural plasticity (or conformational dynam-
ics) of the biocatalyst itself. It has been shown for several
8
,9
CORE subdomain. The enzyme primarily populates two
10
distinct conformations in solution: the open (inactive) and
1
1,12
1
7
1
the closed (active) states.
Interconversion between these
states is associated with large conformational rearrangements of
the ATPlid and AMPbd (Figure 1A,B), and the rate of opening
of the substrate binding subdomains (or alternatively expressed
2
3
the inverse of the lifetime of the closed state (τ )) in the
c
enzymes including dihydrofolate reductase, RNase A, cyclo-
4
5,6
presence of bound substrate is rate-limiting for catalysis (Figure
philin A and Escherichia coli adenylate kinase (AK ) that
eco
5,6
1
D).
Since the definition of kcat is the rate-limiting
conformational dynamics, in fact, can represent the rate-limiting
microscopic step for catalytic turnover. The enzymes described
above display what is referred to as pre-existing equilibria where
the substrate-free states are in equilibrium with “substrate
bound-like” structures that are only scarcely populated. In the
microscopic step for breakdown of the ES complex and
13
generation of product, the rate constant of subdomain
opening in AKeco is equal to k . The molecular mechanism of
cat
14
ATPlid closure follows an “order to disorder to order” or
15
7
“cracking” mechanism where specific helical segments in the
ATPlid unfold with subsequent refolding into the closed
conformation.
conformational selection model, these bound-like structures
harbor the substrate binding activity. To deepen the current
knowledge of enzymatic catalysis, it is crucial to understand the
molecular details that influence conformational dynamics, and
how conformational dynamics affects catalysis. AKeco is an
excellent model system in studies of the linkage between
dynamics and enzymatic catalysis since its function is
dependent on large-scale conformational changes (Figure 1).
Closed-like AKeco conformations are populated already in the
6
,12,17
absence of substrate,
and addition of substrates
Received: April 4, 2012
©
XXXX American Chemical Society
A
dx.doi.org/10.1021/ja3032482 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
directed mutagenesis was performed with the Quickchange approach
(Stratagene) with oligonucleotide primers purchased from Eurofins
MWG Operon. The E170A mutation was verified with DNA
sequencing (Eurofins MWG Operon).
Circular Dichroism. Far ultraviolet CD experiments were
performed on a Jasco J-810 spectropolarimeter. Thermal unfolding
was followed by monitoring the CD signal at 220 nm in a 1 mm
−1
cuvette and a scan rate of 1 deg min . Protein concentrations in the
CD experiments were 7.5 μM in a buffer consisting of 10 mM sodium
phosphate and 50 mM NaCl at pH 7.0. Melting temperatures were
quantified with nonlinear fits (Microcal Origin) of CD-data to a two-
3
0
state transition.
Enzymatic Assays. Enzyme activity in the direction of ADP
31
formation at 25 °C was quantified with a coupled ATPase assay. All
reagents for the assay were purchased from Sigma-Aldrich. Catalytic
parameters were obtained through fits of initial velocities to the
Michaelis−Menten equation. In all conditions, care was taken to
ensure that the amount of coupling enzymes (pyruvate kinase and
lactate dehydrogenase) was not limiting the reactions. AKeco
concentrations were calculated by measurements in triplicate.
Figure 1. Stabilization of the closed AKeco conformation with an
interdomain hydrogen bond suggested from the crystal structure. (A)
1
0
Crystal structure of substrate free open AKeco
. The distance between
ε
the Glu170 side chain hydrogen bond acceptor (O 1 atom, red
sphere) and the Leu58 backbone amide hydrogen bond donor (yellow
sphere) is 16 Å in open AKeco. The ATPlid, AMPbd, and CORE
subdomains are indicated and colored blue, yellow, and gray,
respectively. (B) Structure of closed AKeco in presence of the tight
binding inhibitor Ap5A (khaki). (C) The closed AKeco conformation is
stabilized with a hydrogen bond between Glu170 and Leu58; the
distance (2.3 Å) between hydrogen bond donor and acceptor is
indicated in an expanded and rotated view with respect to (B). The
NMR Spectroscopy. All NMR experiments were acquired on a
Bruker DRX 600 MHz spectrometer equipped with a 5-mm triple
resonance z-gradient cryoprobe. The temperature in all NMR
experiments was 25 °C. Temperature calibration was accomplished
with a temperature probe inserted into the sample compartment of the
cryoprobe. The NMR sample buffer was 30 mM MOPS pH 7.0 and 50
11
mM NaCl with 10% (v/v) 2H O. Protein concentrations ranged
2
between 600 and 900 μM supplemented with 2.5 mM Ap5A (Sigma-
1
6
molecular graphics images were generated with MolMol. (D)
Nucleotide binding mechanism for AKeco. E corresponds to enzyme, S
to substrate, ES to a transient substrate bound equilibrium complex
Aldrich) in case of analysis of closed states. Assignments of E170A in
15
apo and Ap5A saturated states were accomplished with 3D
N
O
NOESY-HSQC experiments using a pulse sequence from the Bruker
library. Hydrogen to deuterium exchange experiments were performed
C
and ES to the active conformation of AKeco. The superscripts “O” and
8
“
C” refer to AKeco in open (inactive) and closed (active)
as described in Rundqvist et al. NMR spectra were processed with
3
2
33
conformations. The equilibrium constants for the initial binding
NMRPipe and visualized in Ansig for Windows.
Molecular Dynamics Simulation. Molecular dynamics simu-
event and the following conformational changes are given by K and
d
3
4
Kconf, respectively. See the Supporting Information for a detailed
description of Kconf. The lifetimes of open and closed states are equal
to τ and τ , respectively, under substrate saturating conditions.
lations were performed using Amber99sb-ildn 8 with explicit
3
5
11
TIP3P 9 solvent on the crystal structure 1AKE.PDB and sodium
counterions to neutralize the system (12 Å solvation shell, cubic box of
o
c
7
7.5 Å with periodic boundary conditions, approximately 47 000
atoms) for 3 × 1 μs stochastic dynamics (time step of 2 fs), and
Particle Mesh Ewald electrostatics. We stress the necessity of the large
solvation shell to prevent AKeco self-interactions with its periodic
images due to conformational motions. Langevin temperature
coupling (300 K, inverse friction constant 0.1 ps) and Berendsen
redistribute pre-existing states by stabilizing the closed
1
2
conformation. Hence, in AK , the same dynamic event
eco
(
i.e., opening/closing) occurs in both substrate free and
substrate bound states. Sophisticated fluorescence techniques
36
can provide detailed single-molecule insight about conforma-
pressure coupling was employed. All simulations are run in Gromacs
18
tional dynamics, but require labeling of specific sites. Solution
state NMR spectroscopy provides label-free approaches to
quantify conformational dynamics in proteins with residue-
specific resolution. In particular, relaxation dispersion experi-
v4.5.4. To calculate water mediation of hydrogen bonds, only water
ε
molecules in the region of 6 Å surrounding the NH of Leu58 and O
of E170 are considered. All possible hydrogen bonds between L58,
E170 and selected waters are calculated using PyMOL using the
37
1
9,20
21
methodology of DSSP and the shortest possible distance (in
ments
can provide information on the structure,
molecules) between L58 and E170 is recorded.
thermodynamics and kinetics of exchange processes in proteins.
The most readily accessible and accurate parameter in NMR
spectroscopy is the chemical shift. Recent developments have
shown that chemical shifts can be used to determine high-
RESULTS AND DISCUSSION
■
Mutational Induced Perturbation of AKeco. The closed
AKeco conformation is stabilized with an interdomain hydrogen
bond suggested from the crystal structure (1AKE.pdb). This
2
2−24
resolution protein structures
conformational dynamics
but also to quantify
1
2,25−28
29
and allostery in proteins.
N
hydrogen bond is donated from the amide hydrogen (H ) of
Here, we show that the open/closed AKeco equilibrium can be
modulated in opposite directions by either removal of one
interdomain hydrogen bond distant from the active site or by
addition of the osmolyte TMAO. Remarkably, modulation of
closed state populations is propagated to proportional
responses in the catalytic parameters kcat and K of AK .
ε
Leu58 in AMPbd to the O side chain oxygens of Glu170 in the
ATPlid (Figure 1C). This interaction has been suggested to
partially explain the cooperative closure of both substrate
binding domains in response to simultaneous binding of ATP
12
and AMP. The hydrogen bond is only present in the closed
state (the distance between donor and acceptor in the open
state is 16 Å Figure 1A), and is thus expected to affect the
open/closed conformational equilibrium. NMR detected
M
eco
MATERIALS AND METHODS
Protein Purification and Mutagenesis. Both wild-type and
■
8
5
hydrogen to deuterium exchange experiments were used to
E170A AK were expressed and purified as described previously.
eco
Protein concentrations were quantified by measuring the absorption at
2
indirectly observe formation of this important hydrogen bond
in closed AKeco in solution (Figure 2). There exist small
80 nm using an extinction coefficient of 10.430 M− cm . Site-
1
−1
B
dx.doi.org/10.1021/ja3032482 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
two interaction interfaces, both of which must be in contact
with Ap5A for AKeco to utilize fully the binding energy. Kconf
describes to which extent both interaction interfaces are
engaged in Ap5A binding once the initial Ap5A:AKeco complex
has been formed.
To access the importance of the interdomain hydrogen bond,
Glu170 was mutated into alanine (E170A) which eliminates the
capacity of the side-chain at residue 170 to accept a hydrogen
bond. E170A is a well folded protein based on the chemical
shift dispersion and narrow distribution of peak intensities in a
1
15
H− N HSQC spectrum (Supporting Information Figure
S1A). On a global scale the chemical shift perturbations in
E170A referenced to the wild type are small (Supporting
Information Figure S1B) showing that the structural
perturbation of the mutant is minimal and that open ground
states are very similar in mutant and wild type. There, however,
exist small but significant differences in chemical shifts for a
limited set of residues and these differences carry important
information on ATPlid dynamics. Figure 3A shows a structural
Figure 2. Indirect observation of the E170-L58 hydrogen bond in
1
15
closed wild-type AKeco. Shown are H− N HSQC expansions
representing protonated reference spectra (blue contours) and the
first spectra after exchange of protonated AKeco into 2H O (red
2
contours). The position of the NH correlation for Leu58 is indicated
in all panels. (A) apo AKeco (open state). (B) AKeco in complex with
Ap5A (closed state). (C) Hydrogen to deuterium exchange kinetics
for the Leu58 amide proton in the closed Ap5A complex, the best
fitted single exponential decay is shown as a blue line. (D) E170A in
complex with Ap5A (closed state).
chemical shift differences between the experimens performed in
Figure 3. Chemical shift perturbations in E170A and TMAO
perturbed AKeco. Significant chemical shifts (Supporting Information
1
2
H O and H O, these differences are likely dependent on
2
2
Figure S1B,C) in E170A (A) and in AK with 0.35 M TMAO (B) are
eco
minute differences in pH values in the samples. In apo AK ,
eco
ε
displayed in red on open AKeco. The O 1 atom of Glu170 is shown as a
N
the hydrogen bond to the H atom of Leu58 is not present in
red sphere on its side chain in panel (A). Chemical shift perturbations
the crystal structure, and consequently the amide hydrogen of
Leu58 exchanges within the dead-time of the experiment (11
min) (Figure 2A). In contrast, this hydrogen atom shows
substantial protection toward exchange in the closed Ap5A
bound state (Figure 2b). Protection against deuterium
exchange is generally considered as indirect evidence of
15
1
are referenced to wild type AKeco according to: δω = 0.2.|δ N| + |δ H|
(ppm). (C) Residues that show significant chemical shift perturbations
in response to ATP binding (adapted from ref 12).
3
8
display of the chemical shift perturbations in E170 compared to
hydrogen bonds. Evaluation of exchange kinetics (using the
wild-type AK . The differences are clustered at positions that
eco
formalism described in ref 39) reveals that the local free energy
12
for the hydrogen bond is 16 kJ mol−1 (Figure 2C) (see
are known to respond to ATPlid dynamics (Figure 3C).
Hydrogen to deuterium exchange experiments suggest that the
hydrogen bond donated by Leu58 is absent in both open (not
shown) and closed E170A conformations (Figure 2D).
The surface exposed mutation does not significantly
influence the stability of either apo (open) or Ap5A bound
Supporting Information and BMRB entry 18685). The
calculation assumes that the exchange occurs in the EX2
regime which previosuly has been shown for residues in the
8
N
AMPbd of AK . Since the H hydrogen atom of Leu58 is
eco
fully exposed to the solvent in the open state, a lower bound of
the free energy difference between open and closed states in the
Ap5A bound complex can be estimated from the stability of the
hydrogen bond. The free energy difference corresponds to an
open/closed equilibrium constant (Kconf in Figure 1D, see the
Supporting Information for a description of Kconf) equal to
(
closed) states as evidenced by marginal (∼0.7 °C) reduction
of melting temperatures (T ) compared to wild-type AK
m
eco
(
Table 1 and Figure 4A,B). Taken together, the NMR and
stability measurements suggest that, overall, the open and
Table 1. Thermal Stabilities of AKeco and E170A
∼
700 in the direction of domain closure. This value should be
compared to the K value for ATP binding which is equal to
a
conf
protein
apo (T , °C)
Ap5A (T , °C)
m
m
1
2
1
.5. By combining the apparent binding constant for Ap5A to
AKeco (0.14 μM ) with the minimal reaction scheme (Figure
D) for Ap5A binding and Kconf equal to 700, the dissociation
b
WT
57.0 ± 0.2
64.5 ± 0.2
65.9 ± 0.2
63.9 ± 0.2
66.5 ± 0.2
1
4
WT + 0.35 M TMAO
E170A
56.9 ± 0.2
56.3 ± 0.2
55.7 ± 0.2
1
constant (K in Figure 1D) for Ap5A was found to be 100 μM.
d
E170A + 0.35 M TMAO
Hence, the effect of the conformational change in AKeco with
respect to Ap5A binding is to increase the binding affinity
almost 3 orders of magnitude (100/0.14). The ATP mimicking
part of Ap5A is sandwiched between residues in the ATPlid and
CORE subdomains (Figure 1B). Hence, Ap5A is bound with
a
The Ap5A concentration was 400 μM (further increase in the
b
concentration did not result in increased Tm values). Errors are
estimated from repeat measurements, the error from the nonlinear
fitting routine is around 0.02 °C.
C
dx.doi.org/10.1021/ja3032482 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 4. Thermal stabilities of wild-type and E170A AKeco. Thermal
stabilities were quantified by following the CD signal at 220 nm as a
function of temperature. Melting points (T ) were quantified by fits of
m
the data to a two state transition (solid lines). The thermal transitions
are 60−90% reversible. In all panels, blue circles corresponds to
regular buffer and red circles corresponds to buffer supplemented with
0.35 M TMAO. (A) apo AKeco; (B) apo E170A; (C) AKeco with Ap5A;
(
D) E170A with Ap5A.
Figure 5. Water-mediated L58/E170 hydrogen bond from MD
simulations. (A) RMSD to the initial closed structure (blue) and
average number of hydrogen bonds connecting L58 and E170 (red) vs
simulation time. During the first 20 ps of the simulation, the closed
conformation opens slightly and water molecules enter to mediate the
interaction between Glu170 and Leu58. (B and C) Details of the
interdomain hydrogen-bond network. The direct hydrogen bond
between the amide hydrogen of Leu58 of AMPbd to O of Glu170 in
the ATPlid, as observed in the crystal structure, can be mediated by
one (B) or two (C) water molecules.
closed structural basins in wild-type and E170A AKeco are very
similar but that ATPlid dynamics is affected.
Inspired by the presence of the L58/E170 hydrogen bond in
closed AK , we performed three 1 μs unrestrained molecular
eco
dynamics (MD) simulations with explicit solvent to probe the
initial events during subdomain opening. Other simulation
ε
40−42
studies have focused on the open to closed transition
and
also at estimation of the free energy difference between open
4
3,44
and closed states.
studies
Since this report (see below) and other
6
,12,17
have shown that AKeco samples closed-like
conformation in absence of substrate, the simulations were
initiated from closed substrate-free AKeco with the L58/E170
hydrogen bond intact. Overall, the three simulations capture
the closed to open transition albeit to different extent. In the
first simulation, AKeco remains in the closed state for almost 1
μs and starts to open toward the end of the simulation (Figure
principle depend on a compactification of the structure due to
the presence of precipitating agents (∼2 M ammonium
11
sulfate ) during the crystallization procedure.
TMAO Induced Perturbation of AKeco. Osmolytes favor
compact protein states through co-solvent exclusion (alter-
48
natively denoted preferential hydration) effects. In ribonu-
clease A, it has been shown that TMAO decreases the native
state flexibility that was induced by the presence of guanidine
5
A). The other two simulations sample the opening event
within 0.1 μs (see Supporting Information Figure S2). In all
simulations, a fast relaxation process that corresponds to a
slight opening of the ATPlid is observed within the first 20 ps.
During this event, the direct hydrogen bond between L58 and
E170 is substituted by an indirect hydrogen bond as one or two
water molecules (Figure 5) enter and satisfy the hydrogen
bonding acceptor and donor capabilities of positions 170 and
49
hydrochloride. Thus, TMAO can affect a protein energy
landscape both at the level of folding/unfolding and native state
dynamics. Therefore, in the case of AK , TMAO may affect
eco
both the unfolding/folding and opening/closing events. These
two processes were approached experimentally by monitoring
effects on Tm for folding/unfolding, and of NMR chemical
shifts for opening/closing. In case of AKeco and E170A apo
states TMAO did not introduce any significant shifts in Tm
(Figure 4A,B and Table 1). The thermal stabilities of the closed
Ap5A bound states were however increased around 2 °C. The
result is interesting since Ap5A binds tightly to AK (K =
5
8. Hence, breathing within the closed structural basin is
accompanied by accommodation of one or two water molecules
that bridge the hydrogen bond between positions 58 and 170.
Still, the presence of early water molecules in MD simulations
should be treated with caution. While MD simulations
stimulate understanding of the dynamics of water hydrogen
bond networks in biomolecular systems and significant effort
has been spent on investigating the role of water mediation,
concerns have been raised that force field artifacts can influence
eco
d
1
4,50
0.14 μM
); thus, any weak nonspecific binding of TMAO to
45
the ATP and AMP sites is occluded by the inhibitor. On the
other hand, TMAO may bind to sites on the protein that is not
occluded by Ap5A. It is thus possible that the stabilizing effect
is due to both a co-solvent exclusion effect and weak binding of
TMAO. For the opening−closing event, it has been shown with
time-resolved FRET experiments that TMAO addition reduces
46
47
hydrogen bonding. On the other hand, it is interesting that in
the crystal structure of closed AKeco (1AKE.PDB) there is no
water molecule bridging this hydrogen bond. This can in
D
dx.doi.org/10.1021/ja3032482 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
the distance between the side chains of residue 142 in the
51
ATPlid and residue 203 in the CORE subdomain. This nicely
illustrates that closed conformations are favored in presence of
1
15
TMAO. A H− N HSQC spectrum of AK in presence 0.35
eco
M TMAO is of high quality (Supporting Information Figure
1
C) and the overall chemical perturbations referenced to AKeco
under regular buffer conditions are relatively small (Supporting
Information Figure 1D). As for E170A, the largest chemical
shift perturbations occur for residues that are sensitive to
ATPlid dynamics. The structural distribution of these
residues overlap with the E170A perturbation pattern (Figure
1
2
3B) which is a first indication that both E170A and osmolyte
perturbations affect the same global exchange process.
Modulation of the Open/Closed Equilibrium. A careful
1
15
inspection of the H− N HSQC spectra of apo wild-type, apo
wild-type with 0.35 M TMAO and apo E170A shows that, for
several residues, the chemical shifts of these three fall onto a
straight line (Figure 6A,B). This behavior indicates that the
chemical shift perturbations in both hydrogen-bond and
TMAO perturbed AKeco are probing the same global exchange
process. Similar linear trends that report on one global process
1
2,29 52
have been observed in several other systems.
. Since FRET
experiments have shown that TMAO affects the open/closed
51
equilibrium by stabilizing closed conformations, we conclude
that both the hydrogen bond and TMAO perturbations affect
the open/closed equilibrium. In all cases, the chemical shifts of
apo AKeco appear in-between those of the two perturbed states.
The linear response is fitted globally for the 27 residues that
follow this pattern in Figure 6C. The linear response to both
mutational and osmolyte driven perturbation is further
substantiating that both perturbations affect the same global
exchange process.
The relative change in populations of the closed state in
hydrogen-bond perturbed and osmolyte perturbed AKeco
referenced to wild-type AKeco under regular buffer conditions
can be estimated from the global analysis of chemical shifts in
Figure 6C. The slope of the linear correlation (δy/δx = 0.45)
corresponds to average chemical shift changes referenced
according to: δy equals the chemical shift changes between
AKeco and E170A, whereas δx equals the chemical shift changes
between 0.35 M TMAO and E170A. These chemical shift
changes place the three states on the open to closed reaction
coordinate; it is however not possible to extract a quantitative
measure on the absolute positions on the reaction coordinate
since the chemical shifts of the open and closed states in
absence of substrate are presently inaccessible. Removal of the
E170/L58 hydrogen bond shifts the equilibrium toward the
open state, whereas TMAO addition shifts the equilibrium
toward the closed state. Since apo AKeco peaks always appear
in-between the E170A and TMAO peaks, the linear response
shows that apo AKeco chemical shifts are linear combinations of
chemical shifts in open and closed states. The global chemical
shift correlation analysis was based upon visual inspection of
Figure 6. Detection and perturbation of the open/closed equilibrium
1
15
in apo AKeco. (A and B) Expansions of H− N HSQC spectra of apo
AKeco (red contours), apo AK in presence of 0.35 M TMAO (purple
eco
contours) and apo E170A (blue contours). Residues Leu5 and Gly122
are shown in (A) and (B), respectively. (C) Global correlation of the
27 chemical shifts that fall onto a straight line in the three states
defined in (A) and (B). The x-axis corresponds to the difference in
chemical shifts between E170A and AKeco with 0.35 M TMAO,
whereas the y-axis corresponds to the difference in chemical shifts
between AK and E170A. Chemical shifts are weighted according to
eco
15
1
12
δω = 0.2·δω N + δω H (ppm) and the slope of the correlation is
equal to 0.45. Residues Leu5 and Gly122 are indicated in blue.
53
NMR spectra. Chemical shift projection analysis is an
alternative approach to identify correlations between chemical
shifts. We applied this approach to our data set and defined an
projection amplitude for residues with cos θ > 0.9 is equal to
0.53 which is strikingly similar to the result from the linear
correlation analysis (0.45). The majority of analyzed residues
place the three states in the same order as deduced from the
linear correlation analysis. Overall, both methods identify very
similar sets of residues as sensitive markers of the opening/
closing reaction (Supporting Information Figure S5), even
though the projection analysis identified a few additional
“activation” vector from E170A toward AKeco with 0.35 M
TMAO (i.e., along the open to closed reaction trajectory, see
Supporting Information Figure S3 for an illustration of the
vector) and computed the projection angle (reported as
cos(θ)) and amplitude (X) of AKeco chemical shifts on this
vector. The resulting projection angles and amplitudes are
shown in Supporting Information Figure S4. The average
E
dx.doi.org/10.1021/ja3032482 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
5
4
residues. Thus, the projection analysis reinforces the conclusion
that the conformational state of AKeco is between those of
E170A and AKeco with 0.35 M TMAO on the open to closed
reaction trajectory.
Menten equation. For the E170A perturbation (decreased
closed population), we observe a 30% increase in the maximum
turnover rate, and for the osmolyte perturbation (increased
closed population), we observe a 20% decrease in the maximum
turnover rate.
These results from both the linear correlation and projection
analysis suggest that AKeco samples the fully closed con-
formation in absence of substrate and, in addition, that the
E170/L58 hydrogen bond is formed and broken during the
exchange between open and closed states. This finding is in
good agreement with previous conclusions from single-
molecule experiments where closed states were inferred from
In Figure 8. the catalytic parameters of the three scenarios are
plotted against relative positions on the open to closed reaction
6
,17
the experimental data in substrate free AKeco.
We find it
notable that perturbations driven by completely different
physical mechanisms (mutation vs osmolyte) affect the same
conformational exchange process in a protein albeit in different
directions.
Catalytic Activities in Response to Mutational/
Osmolyte Perturbation. Next we asked if the mutational
and osmolyte induced perturbation of the open/closed
equilibrium affects the catalytic properties of AK . On the
eco
basis of the chemical shift analysis, the populations of the closed
states are decreased and increased in the hydrogen bond and
TMAO perturbations, respectively. With a reaction coordinate
defined as the trajectory from open to closed conformations,
the three can be placed in the following order: E170A, AKeco,
and AKeco in presence of 0.35 M TMAO. The enzymatic
activities of E170A and osmolyte perturbed AKeco were
quantified with a coupled ATP:ase assay in the direction of
3
1
ADP formation and benchmarked against the wild-type under
regular buffer conditions (Figure 7 and Table 2). All variants
Figure 7. The magnitude of the open/closed equilibrium affects
enzyme kinetics. Enzymatic turnover (scaled with the protein
concentration), derived from a coupled ATP:ase assay with ATP as
variable substrate and AMP constant at 300 μM, is shown together
with fits to Michaelis−Menten kinetics (solid lines) for AK (red),
eco
AK with 0.35 M TMAO (purple), and E170A (blue). Error bars are
eco
derived from triplicate measurements.
Table 2. Enzymatic Parameters of Wild-Type and Perturbed
AKeco
Figure 8. Correlation between enzymatic parameters and chemical
shifts. The relative position on the open to closed reaction coordinate
is from the slope (0.45) in Figure 6C. The x-axis should be interpreted
as: if the distance from E170A to AKeco in the direction of the open
state is 0.45, then the distance from AKeco with TMAO to AKeco in the
direction of the closed state equals “0.55”. The best fitted straight lines
are shown as solid red lines. (A) Correlation between kcat and chemical
shifts. (B) Correlation between KM and chemical shifts. (C)
kcat (s−1)
KM (μM)
k /K (s−1 μM−1
)
cat
M
Wild-type
E170A
360 ± 10
470 ± 10
286 ± 7
63 ± 7
78 ± 6
42 ± 5
5.7 ± 0.7
6.0 ± 0.5
6.8 ± 0.8
0.35 M TMAO
display Michaelis−Menten kinetics and k and K values were
Correlation between k /KM and chemical shifts. The average kcat/
cat
M
cat
quantified by fitting initial reaction velocities to the Michaelis−
KM value is indicated as a dotted line.
F
dx.doi.org/10.1021/ja3032482 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
coordinate. It should be noted that the relative positions cannot
be scaled to obtain a quantitative number since the chemical
shifts of the “true” open and closed substrate free states are
currently inaccessible. It thus seems possible to rationally tune
the maximum turnover rate in AKeco simply by stabilizing or
destabilizing the closed state relative to the open state.
consistent with a small molecule binding to the enzyme for
three reasons: (1) the chemical shift pattern is overlapping with
that of ATP binding (Figure 3B,C), (2) the conformational
equilibrium is shifted toward the closed state with 0.35 M
TMAO (Figure 6C), (3) the T of both substrate free AK
and E170A is not significantly perturbed by 0.35 M TMAO
which is expected for a binding event. Finally, FRET
m
eco
We also observe a strong linear correlation between the KM
values and the population changes in the perturbed AKeco states
experiments have shown that TMAO favors closed AK
eco
51
(
Figure 8B). The KM values in E170A and TMAO
states. It can also not be ruled out that the microscopic
mechanisms of AKeco catalysis are altered in the presence of
TMAO. In summary, for the E170A modification, the
perturbations are 78 and 42 μM, respectively. These values
should be compared to a K of 63 μM for AK under regular
M
eco
buffer conditions. These observations can be rationalized by the
fact that nucleotide binding affinity scale with the closed
populations according to Figure 1D (and acknowledging that
KM is reflecting a nominal ATP binding affinity).
perturbation of kcat and K values is dependent on a population
shift toward the open state of this variant. For the situation of
M
AKeco with 0.35 M TMAO, the effects on kcat and K are likely
M
a result of both a shift toward the closed state and weak residual
binding of TMAO to the enzyme.
However, the KM value can only be treated as a nominal
binding affinity. Thus, it is important to benchmark the ATP
binding affinities for the E170A and TMAO perturbations
CONCLUSIONS
■
12
against the K of the wild-type (53 μM ). To this end, we used
d
Here, we have utilized the full resolution capacity of NMR
spectroscopy to study an enzyme at the level of an individual
NMR titrations to quantify the K values for ATP binding for
d
E170A in regular buffer and wild-type AKeco in presence of 0.35
N
atom (H of Leu58). Moreover, we have shown that hydrogen
M TMAO (Figure 9). Note, that the K values should be
d
bonding to this atom (which is distant from the active site)
directly influences the catalytic properties of AK . MD
eco
simulations indicate that water molecules enable breathing
within the closed AKeco ensemble mediating the hydrogen bond
between Leu58 and Glu170. Since the hydrogen bond affects
the catalytic properties of AK , the involvement of water in
eco
mediating this bond suggests that the water molecules are
directly coupled to AKeco activity. Similarly, structural waters
55
embedded in cavities have been observed in many proteins.
These water molecules often make hydrogen bonds to
otherwise “naked” hydrogen bond donor and acceptor groups
55,56
and can, as a consequence, increase protein stability.
For
AKeco and also other kinases, it is important to remove water
molecules from the active site to prevent nonproductive
hydrolysis of ATP. In AK , this is accomplished with a large-
eco
scale conformational change that together with substrate
binding will generate a dry active site void of water. In light
of our findings, it seems that structural water molecules not
only can influence the stability of proteins, be directly involved
in chemical mechanisms, or cause unwanted side-reactions in
enzymes, but can also take an active part during conformational
changes. Removal of the Leu58 hydrogen bond increases the
Figure 9. NMR detected ATP binding experiments. (A and B)
Residue specific ATP binding isotherms for wild-type AKeco with 0.35
M TMAO. (C and D) Residue specific ATP binding isotherms for
E170A. The solid lines correspond to fits to one-site binding models.
13
catalytic parameters kcat and K by decreasing the population of
the closed and active AKeco conformation. An osmolyte
perturbation acts to increase the same population and results
M
interpreted as “apparent” K values since all substrate bound
d
states will contribute to the binding affinity. The K values were
d
determined for assigned residues in the ATPlid (defined as
residues 113−176) and the reported values are averages over all
in decreased kcat and K values. It is however important to note
that the TMAO effect on the catalytic parameters likely is
M
analyzed residues. For E170A, we observe a K value of 150 ±
dependent on both a shift of the open-closed equilibrium and
d
5
6
44 μM; thus, the hydrogen bond perturbation decreases the
weak nonspecific TMAO binding to AK . We and others
eco
ATP binding affinity which is reflected in an increased KM
value. This result is consistent with a conformational selection
model for ATP binding since the wild-type that is more shifted
toward the closed state (compared to E170A) binds ATP with
higher affinity. In case of AKeco with 0.35 M TMAO, we
observe a K value of 150 ± 16 μM; thus, the decreased K
value cannot be rationalized with an improved ATP binding
activity. In principle, the decreased catalytic activity in presence
of 0.35 M TMAO can depend on weak binding of TMAO to
have previously shown that AKeco catalysis is rate-limited by the
opening of the substrate binding subdomains in the presence of
bound nucleotides. Hence, destabilization of the closed state is
expected to increase the rate of AKeco catalysis. The kinetics of
subdomain opening can alternatively be expressed as the
reciprocal of the lifetime of the closed conformation (= 1/τc).
In this study, we show a strong link between the population of
the closed AKeco state and catalytic activity. In light of the
relatively small changes in catalytic activity between the wild-
type enzyme and the perturbations employed here, it is likely
that the rate-limiting step is subdomain opening also in the
perturbations. Under this assumption, the population changes
are linked to increased and decreased lifetimes of the closed
d
M
AK , and that TMAO is acting as a noncompetitive inhibitor
eco
(
which would explain the reduced ATP binding affinity). This
explanation cannot be ruled out, but it is clear that the chemical
shift perturbation pattern driven by 0.35 M TMAO is not
G
dx.doi.org/10.1021/ja3032482 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
state in the osmolyte and hydrogen bond perturbations,
respectively. Therefore (under the assumption stated above),
our previous observation that AKeco turnover is rate-limited by
conformational dynamics is reinforced and taken to the level
where “rational” modulation of the open/closed equilibrium
results in catalytic parameters that change in a direction that is
predictable from the catalytic model. We find it remarkable that
perturbations driven by either hydrogen bond removal or
osmolyte addition affect the same conformational exchange
process in AKeco. The hydrogen bond and osmolyte
perturbations both affect closed versus open populations albeit
in different directions. This observation suggests that TMAO
perturbations of NMR chemical shifts may open a new avenue
for identification and quantification of conformational dynamics
in proteins. Both kcat and KM scale linearly in response to
perturbation of the open/closed equilibrium, which can be
nicely illustrated by plotting both kcat and KM versus relative
chemical shifts in the wild-type and perturbed states (Figure
Supporting Information Figure S2, RMSD and analysis of L58/
E170 hydrogen bonds for two MD simulations; Supporting
Information Figures S3−S5, chemical shift projection analysis;
hydrogen exchange data for Leu58 in presence of Ap5A and a
N
15
detailed explanation of Kconf; H and N chemical shifts have
been deposited at the BMRB with the following accession
codes: 18683 (wild-type AK ), 18686 (E170A), 18687 (AK
eco
eco
with 0.35 M TMAO). This material is available free of charge
via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Author
magnus.wolf-watz@chem.umu.se
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
8
A,B). It has been shown previously that a conformational
This research was financially supported by the Swedish
Research Council (Grant Number 621-2010-5247), and an
equilibrium within a membrane associated protein substrate
affects the rate of phosphorylation by cAMP-dependent protein
̊
Umea University “Young Researcher Award” to M.W.-W. A.S.
5
7
kinase A. Thus, it appears that dynamics of both substrates
and enzymes needs to be considered in order to picture a
complete “native state” free energy landscape during catalysis. A
pre-existing equilibrium (i.e., in absence of substrate),
corresponding to the opening and closure of AKeco was
inferred from the observation of linear chemical shift trends in
recognizes support by the Impuls- und Vernetzungsfonds of the
Helmholtz-Association and the SimLab NanoMicro (SCC,
Karlsruhe Institute of Technology). The MD simulations were
performed using the resources of bwGRiD.
REFERENCES
1) Radzicka, A.; Wolfenden, R. Science 1995, 267, 90.
2) Boehr, D. D.; McElheny, D.; Dyson, H. J.; Wright, P. E. Science
1
15
■
H− N HSQC NMR experiments. Notably, activity measure-
(
(
ments show that the magnitude of the pre-existing equilibrium
directly influences the catalytic properties of the enzyme. There
exist many reasons to develop enzymes with increased catalytic
activity; our results show that in the case of enzymes where
conformational dynamics is rate-limiting for catalysis in ways
2
(
2
(
006, 313, 1638.
3) Beach, H.; Cole, R.; Gill, M. L.; Loria, J. P. J. Am. Chem. Soc.
005, 127, 9167.
4) Eisenmesser, E. Z.; Millet, O.; Labeikovsky, W.; Korzhnev, D. M.;
similar to AK , an increase (or decrease) of the catalytic
Wolf-Watz, M.; Bosco, D. A.; Skalicky, J. J.; Kay, L. E.; Kern, D. Nature
eco
activity is linked to a proportional change of the Michaelis
2005, 438, 117.
5
8,59
constant. In analogy to enthalpy−entropy compensation
(5) Wolf-Watz, M.; Thai, V.; Henzler-Wildman, K.; Hadjipavlou, G.;
Eisenmesser, E. Z.; Kern, D. Nat. Struct. Mol. Biol. 2004, 11, 945.
(6) Hanson, J. A.; Duderstadt, K.; Watkins, L. P.; Bhattacharyya, S.;
Brokaw, J.; Chu, J. W.; Yang, H. Proc. Natl. Acad. Sci. U.S.A. 2007, 104,
observed for weak interactions in for instance protein folding
6
0
and ligand binding, such systems display k −K compensa-
cat
M
tion upon perturbation (Figure 8C). Similar observations of
1
(
8055.
7) Lange, O. F.; Lakomek, N. A.; Fares, C.; Schroder, G. F.; Walter,
k −K compensation has been observed previously for the
cat
M
61
oxidative repair enzyme AlkB. The k −K compensation
cat
M
K. F.; Becker, S.; Meiler, J.; Grubmuller, H.; Griesinger, C.; de Groot,
B. L. Science 2008, 320, 1471.
also has significance for the evolution of enzymes where the
most important catalytic parameter for an enzyme under
cellular conditions is the specificity constant, k /K . For
(
́
8) Rundqvist, L.; Åden, J.; Sparrman, T.; Wallgren, M.; Olsson, U.;
cat
M
Wolf-Watz, M. Biochemistry 2009, 48, 1911.
enzymes that display a similar co-variation in kcat and K , there
(9) Bae, E.; Phillips, G. N., Jr. Proc. Natl. Acad. Sci. U.S.A. 2006, 103,
2132.
M
exist many alternative routes to obtain a k /K value that is
cat
M
(
10) Mu
̈
ller, C. W.; Schlauderer, G. J.; Reinstein, J.; Schulz, G. E.
optimal for function. For example, in the AKeco case, removal of
one interdomain hydrogen bond (that affects both kcat and KM)
generates an enzyme with a specificity constant that is silent to
the perturbation. Thus, it appears that AKeco has evolved to be
intrinsically robust with respect to the k /K ratio. As a
Structure 1996, 4, 147.
(
(
(
11) Mu
12) Åden
13) Fersht, A. Structure and mechanism in protein science. A guide to
̈
ller, C. W.; Schulz, G. E. J. Mol. Biol. 1992, 224, 159.
́
, J.; Wolf-Watz, M. J. Am. Chem. Soc. 2007, 129, 14003.
cat
M
enzyme catalysis and protein folding, W.H. Freeman and Company: New
York, 2000.
concluding remark, our study highlights the tight and complex
interplay between native state enzyme dynamics and catalytic
properties.
(14) Olsson, U.; Wolf-Watz, M. Nat. Commun. 2010, DOI: 10.1038/
ncomms1106.
Abbreviations. AK , Adenylate kinase from Escherichia
(15) Whitford, P. C.; Miyashita, O.; Levy, Y.; Onuchic, J. N. J. Mol.
Biol. 2007, 366, 1661.
eco
coli; E170A, AKeco E170A mutation; HSQC, heteronuclear
single quantum coherence; NMR, Nuclear magnetic resonance;
TMAO, trimethylamine N-oxide.
(
5
(
16) Koradi, R.; Billeter, M.; Wu
1.
17) 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.
18) Gambin, Y.; Schug, A.; Lemke, E. A.; Lavinder, J. J.; Ferreon, A.
̈
thrich, K. J. Mol. Graphics 1996, 14,
ASSOCIATED CONTENT
Supporting Information
■
̈
*
S
(
1
15
Supporting Information Figure S1, H− N HSQC spectra of
C. M.; Magliery, T. J.; Onuchic, J. N.; Deniz, A. A. Proc. Natl. Acad. Sci.
U.S.A. 2009, 106, 10153.
the three variants AK , E170A and AK with 0.35 M TMAO;
eco
eco
H
dx.doi.org/10.1021/ja3032482 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(
1
(
19) Loria, J. P.; Rance, M.; Palmer, A. G. J. Am. Chem. Soc. 1999,
21, 2331.
20) Tollinger, M.; Skrynnikov, N. R.; Mulder, F. A.; Forman-Kay, J.
D.; Kay, L. E. J. Am. Chem. Soc. 2001, 123, 11341.
21) Bouvignies, G.; Vallurupalli, P.; Hansen, D. F.; Correia, B. E.;
(56) Takano, K.; Yamagata, Y.; Yutani, K. Protein Eng. 2003, 16, 5.
(57) Masterson, L. R.; Yu, T.; Shi, L.; Wang, Y.; Gustavsson, M.;
Mueller, M. M.; Veglia, G. J. Mol. Biol. 2011, 412, 155.
(58) Liu, L.; Guo, Q. X. Chem. Rev. 2001, 101, 673.
(
(59) Dunitz, J. D. Chem. Biol. 1995, 2, 709.
Lange, O.; Bah, A.; Vernon, R. M.; Dahlquist, F. W.; Baker, D.; Kay, L.
(60) Schug, A.; Onuchic, J. N. Curr. Opin. Pharmacol. 2010, 10, 709.
E. Nature 2011, 477, 111.
(61) Yu, B.; Hunt, J. F. Proc. Natl. Acad. Sci. U.S.A. 2009, 106, 14315.
(
22) Cavalli, A.; Salvatella, X.; Dobson, C. M.; Vendruscolo, M. Proc.
Natl. Acad. Sci. U.S.A. 2007, 104, 9615.
23) Shen, Y.; Lange, O.; Delaglio, F.; Rossi, P.; Aramini, J. M.; Liu,
(
G.; Eletsky, A.; Wu, Y.; Singarapu, K. K.; Lemak, A.; Ignatchenko, A.;
Arrowsmith, C. H.; Szyperski, T.; Montelione, G. T.; Baker, D.; Bax, A.
Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 4685.
(
(
24) Sborgi, L.; Verma, A.; Munoz, V.; de Alba, E. PLoS One 2011, 6.
25) Berjanskii, M. V.; Wishart, D. S. J. Am. Chem. Soc. 2005, 127,
1
(
4970.
26) Berjanskii, M. V.; Wishart, D. S. Nucleic Acids Res. 2007, 35,
W531.
(
(
(
27) Berjanskii, M. V.; Wishart, D. S. J. Biomol. NMR 2008, 40, 31.
28) Berjanskii, M.; Wishart, D. S. Nat. Protoc. 2006, 1, 683.
29) Selvaratnam, R.; Chowdhury, S.; VanSchouwen, B.; Melacini, G.
Proc. Natl. Acad. Sci. U.S.A. 2011, 108, 6133.
30) Consalvi, V.; Chiaraluce, R.; Giangiacomo, L.; Scandurra, R.;
Christova, P.; Karshikoff, A.; Knapp, S.; Ladenstein, R. Protein Eng.
000, 13, 501.
(
2
(
31) Rhoads, D. G.; Lowenstein, J. M. J. Biol. Chem. 1968, 243, 3963.
(
32) Delaglio, F.; Grzesiek, S.; Vuister, G. W.; Zhu, G.; Pfeifer, J.;
Bax, A. J. Biomol. NMR 1995, 6, 277.
33) Helgstrand, M.; Kraulis, P.; Allard, P.; Har
000, 18, 329.
34) Lindorff-Larsen, K.; Piana, S.; Palmo, K.; Maragakis, P.; Klepeis,
J. L.; Dror, R. O.; Shaw, D. E. Proteins: Struct., Funct., Bioinf. 2010, 78,
950.
35) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R.
W.; Klein, M. L. J. Chem. Phys. 1983, 79, 926.
36) Van der Spoel, D.; Lindahl, E.; Hess, B.; Groenhof, G.; Mark, A.
E.; Berendsen, H. J. C. J. Comput. Chem. 2005, 26, 1701.
(
̈
d, T. J. Biomol. NMR
2
(
1
(
(
(
37) Kabsch, W.; Sander, C. Biopolymers 1983, 22, 2577.
(
38) Englander, S. W.; Mayne, L.; Bai, Y.; Sosnick, T. R. Protein Sci.
1
(
5
997, 6, 1101.
39) Englander, S. W.; Kallenbach, N. R. Q. Rev. Biophys. 1984, 16,
21.
(
(
40) Pontiggia, F.; Zen, A.; Micheletti, C. Biophys. J. 2008, 95, 5901.
41) Brokaw, J. B.; Chu, J. W. Biophys. J. 2010, 99, 3420.
(
42) Snow, C.; Qi, G. Y.; Hayward, S. Proteins: Struct., Funct., Bioinf.
2
(
2
(
007, 67, 325.
43) Potoyan, D. A.; Zhuravlev, P. I.; Papoian, G. A. J, Phys, Chem, B
012, 116, 1709.
44) Shehu, A.; Kavraki, L. E.; Clementi, C. Proteins: Struct., Funct.,
Bioinf. 2009, 76, 837.
45) Nygaard, R.; Valentin-Hansen, L.; Mokrosinski, J.; Frimurer, T.
M.; Schwartz, T. W. J. Biol. Chem. 2010, 285, 19625.
46) Levy, Y.; Onuchic, J. N. Annu. Rev. Biophys. Biomol. Struct. 2006,
5, 389.
47) Cerutti, D. S.; Freddolino, P. L.; Duke, R. E., Jr.; Case, D. A. J.
Phys. Chem. B 2010, 114, 12811.
(
(
3
(
(48) Lin, T. Y.; Timasheff, S. N. Biochemistry 1994, 33, 12695.
(49) Doan-Nguyen, V.; Loria, J. P. Protein Sci. 2007, 16, 20.
(50) Schrank, T. P.; Bolen, D. W.; Hilser, V. J. Proc. Natl. Acad. Sci.
U.S.A. 2009, 106, 16984.
51) Nagarajan, S.; Amir, D.; Grupi, A.; Goldenberg, D. P.; Minton,
A. P.; Haas, E. Biophys. J. 2011, 100, 2991.
52) Volkman, B. F.; Lipson, D.; Wemmer, D. E.; Kern, D. Science
001, 291, 2429.
53) Selvaratnam, R.; VanSchouwen, B.; Fogolari, F.; Mazhab-Jafari,
M. T.; Das, R.; Melacini, G. Biophys. J. 2012, 102, 630.
(
(
2
(
(
54) Michaelis, L.; Menten, M. L. Biochem. Z. 1913, 49, 333.
(
55) Park, S.; Saven, J. G. Proteins: Struct., Funct., Bioinf. 2005, 60,
450.
I
dx.doi.org/10.1021/ja3032482 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX