Cosolvent Effects
on Protein Stability
Annu. Rev. Phys. Chem. 2013.64:273-293. Downloaded from www.annualreviews.org
by RENSSELAER POLYTECHNIC INSTITUTE on 04/05/13. For personal use only.
Deepak R. Canchi1 and Angel E. Garcı́a2
1
Department of Chemical and Biological Engineering, 2 Department of Physics, Applied Physics
and Astronomy and Center for Biotechnology and Inter-Disciplinary Studies, Rensselaer
Polytechnic Institute, Troy, New York 12180; email:
[email protected]
Annu. Rev. Phys. Chem. 2013. 64:273–93
Keywords
First published online as a Review in Advance on
January 4, 2013
urea, TMAO, preferential interaction, molecular dynamics simulations,
protein folding
The Annual Review of Physical Chemistry is online at
physchem.annualreviews.org
This article’s doi:
10.1146/annurev-physchem-040412-110156
c 2013 by Annual Reviews.
Copyright
All rights reserved
Abstract
Proteins are marginally stable, and the folding/unfolding equilibrium of proteins in aqueous solution can easily be altered by the addition of small organic
molecules known as cosolvents. Cosolvents that shift the equilibrium toward
the unfolded ensemble are termed denaturants, whereas those that favor the
folded ensemble are known as protecting osmolytes. Urea is a widely used
denaturant in protein folding studies, and the molecular mechanism of its
action has been vigorously debated in the literature. Here we review recent
experimental as well as computational studies that show an emerging consensus in this problem. Urea has been shown to denature proteins through
a direct mechanism, by interacting favorably with the peptide backbone as
well as the amino acid side chains. In contrast, the molecular mechanism by
which the naturally occurring protecting osmolyte trimethylamine N-oxide
(TMAO) stabilizes proteins is not clear. Recent studies have established the
strong interaction of TMAO with water. Detailed molecular simulations,
when used with force fields that incorporate these interactions, can provide
insight into this problem. We present the development of a model for TMAO
that is consistent with experimental observations and that provides physical
insight into the role of cosolvent-cosolvent interaction in determining its
preferential interaction with proteins.
273
1. INTRODUCTION
Proteins are biological macromolecules that carry out a range of essential functions in a cell.
Structurally, they are heteropolymers of the 20 naturally occurring α-amino acids. The sequence
in which the amino acids are joined together, by the peptide bond, defines the primary structure of
the protein. Most proteins adopt a well-defined, three-dimensional structure in solution, called the
folded or native state, which is essential to enable their biological functions (Figure 1). Anfinsen (1)
a
Amino
H
N
+H
C
+NH
3
O
CH
C
O
R2
C
CH
NH
R1
O–
H
NH
C
CH
O
R3
O
R4
C
CH
O–
C
NH
O
R
N terminus
Polypeptide chain
Amino acid
C terminus
Primary structure
Tertiary (folded) structure
b
Urea
Trimethylamine oxide
c
Urea:TMAO
Urea
Water
F
F
U
F
U
Urea
U
TMAO
d
0
ΔG
ΔG
Annu. Rev. Phys. Chem. 2013.64:273-293. Downloaded from www.annualreviews.org
by RENSSELAER POLYTECHNIC INSTITUTE on 04/05/13. For personal use only.
Carboxyl
H
0
2
ΔG([C ]) = ΔG(0) + m[C]
0
ΔG([C]) = ΔG(0) – m[C]
4
6
[Urea] (M)
0
1
2
3
4
[TMAO] (M)
Figure 1
Protein structure, folding equilibrium, and cosolvents. (a) Levels of protein structure. The repeating unit -C-O-N-Cα - makes up the
backbone of the protein, and Ri represent the various side chains. (b) Molecular structure of urea (denaturant) and TMAO (protecting
osmolyte). Carbon is shown in cyan, nitrogen in blue, oxygen in red, and hydrogen in white. (c) The folding equilibrium of the protein
in water is shifted toward the unfolded (U) ensemble in urea. The addition of urea and TMAO in a concentration ratio of 2:1 leaves the
folding (F) equilibrium unaffected. (d ) The free energy of protein unfolding depends linearly on the concentration of the cosolvent.
The slope of the fit is known as the m-value.
274
Canchi
·
Garcı́a
Annu. Rev. Phys. Chem. 2013.64:273-293. Downloaded from www.annualreviews.org
by RENSSELAER POLYTECHNIC INSTITUTE on 04/05/13. For personal use only.
demonstrated that all the information required for a protein to fold to its native state is determined
by its amino acid sequence. Proteins fold to a native structure and on short timescales (typically
milliseconds), despite having a vast conformational space available to them. This is explained by
invoking a funnel-like free energy landscape for the protein in which the folded state lies at the
minimum and by noting that the sequences of naturally occurring proteins are minimally frustrated
as opposed to random heteropolymers (2, 3).
A protein in solution exists in conformational equilibrium with an ensemble of unfolded
states, with the folded ensemble being favored at ambient conditions (Figure 1). The equilibrium
between the folded and the unfolded ensembles can be perturbed by changing the thermodynamic
state of the system (temperature, pressure, pH) or by changing the composition by the addition
of cosolvents to the solution (4–7). The process of shifting the conformational equilibrium
toward the unfolded ensemble is known as denaturation. Denaturation is an important process in
biochemical studies as thermodynamic functions, such as the free energy, enthalpy, and entropy,
which provide information about the stability of the folded state, can be obtained only by perturbing the equilibrium to populate the unfolded (denatured) states (4). The population in different
ensembles can be obtained by monitoring quantities such as spectral properties, heat capacity,
and enzyme activity, from which the equilibrium constant and other thermodynamics quantities
can be derived. Many proteins undergo a sharp, cooperative transition from the native to the
denatured state upon an increase in temperature, leading to the characteristic sigmoidal curve for
the experimental observable (4). Such transitions are treated in a two-state model, with the validity
of the assumption judged by comparing the van’t Hoff enthalpy with the enthalpy measured by
calorimetry.
Proteins are also, counterintuitively, denatured by increasing the pressure and unfold with
a decrease in partial molar volume (6, 8–10). It turns out that the packing of the folded state
is not optimal, and the volume of the water-swollen unfolded state is smaller than the folded
state owing to the hydration of the interior hydrophobic residues (11–13). Phase diagrams for the
pressure-temperature behavior of the folding-unfolding transition of proteins have been obtained,
from experiments as well as molecular simulations (14–18), providing a unified thermodynamic
description for pressure, heat, and cold denaturation.
In this review, we describe how the addition of small organic molecules, known as cosolvents,
affects the folding equilibrium of proteins (7). These molecules are termed cosolvents as they occupy a significant fraction of the solution volume. Remarkably, the equilibrium can be favored in
either direction, depending on the identity of the cosolvent (Figure 1b,c). Cosolvents such as urea
and guanidinium chloride that favor the unfolded states are known as denaturants. Conversely,
compounds such as trimethylamine N-oxide (TMAO), glycine, betaine, glycerol, and sugars stabilize the folded state of proteins and are known as protecting osmolytes (7, 19–21). Intriguingly,
TMAO can counteract the denaturing effect of urea on proteins, typically in a 2:1 concentration
ratio of urea and TMAO (19, 22). We focus on the effects of urea and TMAO on protein stability
and show how an interplay of molecular dynamics (MD) simulations and experiments can prove
successful in uncovering the molecular basis of these phenomena.
Cosolvent: small,
neutral organic
molecules that can
modulate protein
stability
van’t Hoff enthalpy:
enthalpy obtained by
analysis of the
temperature
dependence of the
equilibrium constant
TMAO:
trimethylamine
N-oxide
Molecular dynamics
(MD): computational
technique to study the
microscopic motions
of atoms and
molecules; molecular
trajectories are
obtained by
numerically
integrating Newton’s
equations of motion
Linear extrapolation
model (LEM): used
to model the
dependence of protein
stability on cosolvent
concentration
2. PROTEIN DENATURATION BY UREA
The ability of urea to denature proteins has been long known (23) and is used ubiquitously in
protein folding studies. Experimentally, it is observed that the free energy of protein unfolding
decreases linearly with urea concentration. This observation is widely used to estimate the stability
of proteins in water via the linear extrapolation model (LEM); i.e., GUrea = GWater − m[C]
(24, 25) (Figure 1d ). The slope of the linear fit, known as the m-value, is a measure of the response
www.annualreviews.org • Cosolvent Effects on Protein Stability
275
Annu. Rev. Phys. Chem. 2013.64:273-293. Downloaded from www.annualreviews.org
by RENSSELAER POLYTECHNIC INSTITUTE on 04/05/13. For personal use only.
Hydrophobic effect:
the tendency of
nonpolar molecules to
aggregate in aqueous
solution and exclude
water
Dispersion
interaction: weak,
attractive noncovalent
interaction between
electrically neutral
molecules, which
includes dipole-dipole,
dipole-induced dipole,
and London forces
276
of protein stability to the addition of urea. It is of interest to understand the molecular mechanism
by which urea denatures proteins, given its importance in protein folding. This question has
attracted much debate over the past 50 years, the challenge being that protein-urea interactions
are weak and large concentrations are required to denature proteins.
2.1. Suggested Mechanisms
There are two basic schools of thought in explaining the denaturing effect of urea on proteins.
The indirect mechanism postulates that urea causes the denaturation of proteins by the alteration
of water structure, a view that emerged from transfer experiments showing that hydrocarbons are
more soluble in aqueous urea than water (26). The implication here is that the addition of urea
alters the structure of water to enable the ready solvation of alkanes by weakening the hydrophobic
effect, which is a major driving force for protein folding (27). This view has been questioned by
recent experimental and computational studies demonstrating that urea integrates well into the
hydrogen-bonding network of water without affecting water’s spatial distribution around itself,
while showing a minimal tendency for self-aggregation (28–34). To our knowledge, there have
been no experimental studies that provide convincing evidence linking the denaturing effect of
urea to its effect on water structure.
Conversely, we have the direct mechanism, in which urea denatures proteins owing to its local
interaction with the protein, rather than owing to differences in the global properties of aqueous
urea over water. Proteins are chemically heterogeneous, consisting of the peptide backbone and
amino acid side chains, which can be polar, apolar, or charged. Within the direct mechanism
model, the nature and strength of urea’s interactions with the various protein moeities have been
extensively studied. An enduring explanation for this phenomenon has been that urea unfolds
proteins by forming hydrogen bonds with the protein backbone (21, 35). The attractiveness of
this idea is easy to see: Urea is chemically similar to the peptide backbone, and it is believed that
urea competes with the intrabackbone hydrogen bonds that stabilize protein secondary structure
motifs, such as α-helices and β-sheets, to cause protein unfolding. However, urea can also have
favorable interaction with the various amino acid side chains in the protein (34, 36–38). Quantifying
the interactions of urea with individual amino acids and the contribution of these interactions to
the overall thermodynamics of the phenomenon will provide a clear picture of the molecular basis
of chemical denaturation. MD simulations have been used to examine the nature of protein-urea
interactions, i.e., by studying the contribution of electrostatic and dispersion interactions to the
energetics of unfolding to provide insight into the groups interacting with urea. If the hydrogen
bonding of urea to the peptide backbone drives unfolding, then the electrostatic interactions
must be dominant. On the contrary, the dominance of dispersion interactions points to a weak
nonspecific interaction of urea with the entire protein surface, including the apolar groups.
Examining the process from a thermodynamic viewpoint by studying the enthalpy-entropy
contributions to the free energy can help distinguish between the indirect and direct mechanisms.
The free energy of unfolding decreases linearly with urea concentration (24, 25), and an enthalpic
origin of this behavior is indicative of the direct mechanism. If the entropic contribution
dominates, then nonlocal effects such as entropic gain due to the displacement of water by larger
urea molecules must be accounted for (39). An aspect that has not received much attention is
the study of the thermodynamics of urea denaturation as a function of temperature and pressure
(13). Urea is used in high-pressure denaturation experiments to lower the pressure at which the
protein unfolds, but it is not clear if the different modes of protein denaturation are additive or
interact in a synergistic manner.
Canchi
·
Garcı́a
Annu. Rev. Phys. Chem. 2013.64:273-293. Downloaded from www.annualreviews.org
by RENSSELAER POLYTECHNIC INSTITUTE on 04/05/13. For personal use only.
2.2. Insights from Molecular Dynamics Simulations
Urea denaturation has been extensively studied using MD simulations of both model systems and
proteins. Model systems are chosen such that the effects of charge and polarity can be separated
from purely hydrophobic effects. Studies of small hydrophobic solutes in urea solutions have led
to conflicting results regarding the weakening of the hydrophobic effect owing to the addition
of urea (40–44). However, the effect of urea has been clearly observed in simulations of model
hydrophobic systems at a larger length scale. England et al. (45) showed that the attraction between
hydrophobic plates is reduced in urea because of the stabilization of the liquid phase between the
plates against dewetting. A purely hydrophobic polymer was shown to unfold in urea owing to the
formation of enthalpically favorable dispersion interactions of the denaturant with the polymer
(46). Similarly, it has been found that urea accumulates around and in the interior of carbon
nanotubes (47, 48). These studies show that urea can modulate the hydrophobic effect by making
energetically favorable contacts with the purely hydrophobic groups. Urea has also been found to
interact favorably with charged solutes through hydrogen bonding and to disrupt ion pairs (40, 44).
Most reported simulation studies of proteins in aqueous urea have sought to understand the
phenomenon in a mechanistic manner, i.e., unfold the protein starting from the native state and
calculate observables based on a few simulation trajectories (33, 44, 49–53). Some studies attributed
the denaturation to both direct and indirect mechanisms (50, 51), whereas others emphasized the
role of hydrogen bonding between urea and the protein backbone (44, 54, 55). Simulations of
individual amino acids in aqueous urea have shown that urea has a favorable interaction with almost
all amino acids, leading to its preferential binding to amino acids over water (34, 36). Large-scale
simulations of lysozyme in urea demonstrated the importance of urea’s dispersion interactions
in facilitating protein unfolding (33). The aforementioned simulation studies provided valuable
insight into various modes of urea’s interaction with proteins, especially by highlighting the role
of protein-urea dispersion interactions. However, these studies do not address the balance of the
driving forces for denaturation, which can be obtained only from simulations of the protein folding
equilibrium in the presence of urea (56, 57).
The choice of the force fields used to model the solvent environment needs particular
attention. To simulate cosolvent effects on protein stability using MD, one must address the
balance of interactions among the various species in solution. It is necessary that the force fields
employed adequately reproduce solution thermodynamic data for the cosolvent-water system
before the interaction with the protein is considered. Solution thermodynamic data such as partial
molar volumes, activity coefficients, and osmotic coefficients are reflections of the solute-solute,
solute-solvent, and solvent-solvent interactions in the solution. Smith and coworkers (58–61) have
pioneered the parameterizing of solute/cosolvents to reproduce bulk solution thermodynamic
properties using the Kirkwood-Buff (KB) theory for solutions, a statistical mechanical theory for
multicomponent solutions that relates the integrals of pair correlation functions of the species to
bulk thermodynamic measurements (62, 63). The KBFF model (58), which is the model for urea
developed using the KB theory, and the OPLS model (64) are commonly used for simulations
of urea-water mixtures. In an insightful study, Horinek & Netz (34) showed that unlike the
KBFF model, the OPLS model displays strong urea-urea attraction, leading to a deviation from
experimental activity data. The nonideality in the OPLS model results in erroneous contributions
from direct and indirect effects when considering the interaction of urea with peptides and
proteins, whereas the indirect contribution in the KBFF model is negligible. In the following
sections, we introduce experimental methods and models used to study and interpret the effect of
urea (and other cosolvents) on protein stability before bridging computational and experimental
results.
www.annualreviews.org • Cosolvent Effects on Protein Stability
Force field: specifies
the functional form
and the numerical
parameters for
interaction among
molecules in an MD
simulation
277
2.3. Transfer Model
Förster resonance
energy transfer
(FRET): provides a
measurement of the
distance between two
fluorophores
Annu. Rev. Phys. Chem. 2013.64:273-293. Downloaded from www.annualreviews.org
by RENSSELAER POLYTECHNIC INSTITUTE on 04/05/13. For personal use only.
Small-angle X-ray
scattering: provides
information about the
shape and size of
macromolecules
Equilibrium dialysis:
technique to study the
interaction of
cosolvents with a
macromolecule by
measuring the
cosolvent
concentration in two
solutions separated by
a semipermeable
membrane that does
not allow the
macromolecule to pass
Osmometry:
technique to
determine
protein-cosolvent
interaction by
measuring changes in
vapor pressure of the
water owing to the
addition of protein
and/or cosolvent
One popular model to study the influence of cosolvents on protein stability is the transfer model
proposed by Tanford (65). In this model, the folding/unfolding of the protein and the process of
the transfer of folded and unfolded states from water to aqueous urea solution are tied together in
a thermodynamic cycle. According to the transfer model, the free energy of protein unfolding in
Water
) through
a 1-M urea solution (GUrea
1M ) is related to the free energy of unfolding in water (G
Urea
Water
tr
+
ni αi δgi .
(1)
G1M = G
i
Here α i is the fractional change in the solvent-accessible surface area of group i upon unfolding, δgitr
is the experimentally measured free energy of transfer of group i from water to a 1-M urea solution,
and ni is the number of groups of type i present in the protein (22, 35). In this model, it is assumed
that the interaction of various groups with urea contributes in an additive manner to unfolding.
Solubility measurements are used to determine δg tr for amino acids and model compounds (22),
and the exposed surface area in the unfolded ensemble, and thereby α i , is determined using polymer
models (35).
The contribution of each side-chain group is obtained by subtracting the δg tr of glycine from
the δg tr of the corresponding amino acid. The transfer model can be related to the m-value by
invoking the LEM; i.e., i ni αi δgitr = −m. Using this approach, Bolen and coworkers (21, 35,
66) successfully predicted m-values for a large number of proteins. Decomposing the m-value
into group contributions, they suggested that the dominant contribution to the free energy of
unfolding in urea arises from favorable interactions of urea with the peptide backbone, and the
total contribution of the interactions between urea and the side chains may even be unfavorable.
The application of this model to protecting osmolytes suggested that the increased stability of
proteins in osmolytes results from unfavorable backbone-osmolyte interactions (21).
Single-molecule Förster resonance energy transfer (FRET) experiments as well as simulations
have shown that the conformations in the denatured ensemble expose more surface area as the
concentration of the denaturant increases (56, 67–69). The transfer model, as described above,
assumes an unfolded ensemble that is independent of denaturant concentration. Haran and colleagues (70, 71) proposed a modification to the transfer model in which folding is preceded by a
coil–to–molten globule collapse transition and were able to demonstrate that denaturants modulate folding by affecting the collapse transition of proteins. In this model, the free energy of
collapse from a denatured state to a collapsed state has a linear dependence on denaturant concentration, with the slope similar to the m-value from LEM analysis. However, the dependence
of the unfolded ensemble on denaturant concentration has not been fully accepted because of
disagreement between FRET and small-angle X-ray scattering data, in which no dependence of
the scattering profile on the denaturant is observed (72).
2.4. Preferential Interaction Coefficients
The interaction of cosolvents with proteins or other biomolecules can be quantified by measuring
the preferential interaction coefficient, defined as
∂μ2
∂m3
Ŵ=−
=
,
(2)
∂μ3 m2 ,T ,P
∂m2 μ3 ,T ,P
where μ is the chemical potential, m is the concentration, and the subscripts 1, 2, and 3 indicate
water, the protein, and the cosolvent, respectively (73). Preferential interaction is measured using
either equilibrium dialysis (74) or, more recently, vapor pressure osmometry (20). It represents
278
Canchi
·
Garcı́a
a
b
Bulk domain
c
Urea
2
Γ>0
ΔΓ > 0
6
Local domain
Γ<0
ΔΓ < 0
0
Γ (r)
4
Γ (r)
TMAO
2
–2
r
0
–4
–2
Water
Cosolvent
0
0.5
Annu. Rev. Phys. Chem. 2013.64:273-293. Downloaded from www.annualreviews.org
by RENSSELAER POLYTECHNIC INSTITUTE on 04/05/13. For personal use only.
r (nm)
1
0
Folded ensemble
Unfolded ensemble
0.5
1
r (nm)
Figure 2
Preferential interaction, denaturants and osmolytes. (a) Schematic illustrating the interaction of a protein with water and a cosolvent.
The preferential interaction is calculated from molecular dynamics simulations using Ŵ = N clocal
s −(
N cbulk
local
s
bulk )N w ,
Nw
where Ncs and Nw
denote the number of cosolvent and water molecules, respectively. The quantity is calculated as a function of the boundary of the local
domain of the protein, r. (b) The preferential interaction of a protein with urea. Both Ŵ and Ŵ = Ŵ Unfold − Ŵ Fold are positive,
signifying the accumulation of urea in the local domain of the protein. (c) The preferential interaction of a protein with TMAO. Both Ŵ
and Ŵ are negative, signifying the exclusion of TMAO from the local domain of the protein.
the change in the chemical potential of a protein in response to the addition of a cosolvent and
can also be expressed as the change in the cosolvent concentration to maintain constant chemical
potential when a protein is added to the solution. This has been interpreted using a two-domain
model as the difference in the cosolvent concentration in the local domain of the protein and the
bulk solution (75, 76); i.e.,
bulk
N3
local
N
,
(3)
Ŵ = N 3local −
1
N 1bulk
where N denotes the number of molecules (Figure 2). Denaturants such as urea show a positive
value of Ŵ, implying an accumulation of the cosolvent in the vicinity of the protein owing to a
net favorable interaction. Conversely, protecting osmolytes show negative values of Ŵ; i.e., they
are excluded from the local domain of the protein because of net unfavorable interactions with
the protein surface (7). The above expression lends itself to the calculation of the preferential
interaction directly from all-atom MD simulations (77). The preferential interaction is calculated
as a function of the distance from the protein surface, and a suitable cutoff distance is applied to
determine the value of Ŵ.
The difference in the preferential interaction of the cosolvent with the protein in its unfolded
and folded states can be related to the free energy of protein unfolding in the presence of the
cosolvent using the Wyman linkage relation (73),
∂ ln K
= Ŵ = Ŵ Unfold − Ŵ Fold .
(4)
∂ ln a 3 m2
Here K is the equilibrium constant for the folding reaction Fold ⇋ Unfold, and a3 is the activity of
the cosolvent. Simplifying the above expression by using the LEM for the free energy of unfolding
(Figure 1d ), we obtain a relation between the m-value and preferential interaction,
m = RT
|Ŵ|
,
[C]
(5)
www.annualreviews.org • Cosolvent Effects on Protein Stability
279
0M
1.9 M
3.8 M
5.8 M
0.8
0.6
0.4
0.2
300
400
500
600
Temperature (K)
Amber94
14
0.4
0.2
0
300
8
1.9 M
3.8 M
5.8 M
10
8
300
4
400
Folded
Unfolded
0
0
50
100
150
Number of residues
Amber94
Amber99sb
–10
–15
–5
Coulomb
–10
LJ
–20
–25
Coulomb
–15
300
400
Temperature (K)
–30
LJ
–20
200
Temperature (K)
5
5
2
500
Experimental
data
Amber99sb
Experiment
Amber94
0
6
Unfolded
500
d
10
12
400
10
Temperature (K)
Amber99sb
12
0.6
ΔEu
Helical residues
Annu. Rev. Phys. Chem. 2013.64:273-293. Downloaded from www.annualreviews.org
by RENSSELAER POLYTECHNIC INSTITUTE on 04/05/13. For personal use only.
c
0.8
ΔEu
0
Amber99sb
1
m-value/kJ (mol M) –1
Amber94
1
Fraction folded
b
Fraction folded
a
2
4
[Urea] (M)
6
2
4
[Urea] (M)
Figure 3
Folding equilibrium of the Trp-cage miniprotein in urea using replica exchange molecular dynamics (56, 57). (a) Structure of the
20-residue Trp-cage miniprotein. Tryptophan is shown in cyan. (b) Melting curves of Trp-cage in various concentrations of urea,
shown for the Amber94 and Amber99sb models. The m-values obtained from the simulations agree well with the experimental
measurement for Trp-cage, shown along with experimental m-values for larger globular proteins. Panel b adapted with permission from
Reference 56. Copyright 2010, American Chemical Society. (c) The average number of helical residues in protein as a function of
temperature. Unlike the Amber99sb model, the unfolded ensemble in Amber94 shows significant helical content. (d ) Change in
protein-urea interaction energy upon unfolding. For both the models, the Lennard-Jones (LJ) interaction makes a significant
contribution to the total change. Panels c and d adapted with permission from Reference 57. Copyright 2011, Elsevier.
Replica exchange
molecular dynamics
(REMD): an
algorithm used to
enhance the sampling
rugged energy
landscapes of complex
systems such as
protein folding
280
where [C] is the molar concentration of the cosolvent. This expression provides a molecular
interpretation for the m-value in terms of the equilibrium distribution of the cosolvent around
the protein. Ŵ is positive for denaturants, implying a greater preferential interaction between
the protein and the cosolvent in the unfolded ensemble. Alternatively, osmolytes are expected
to be excluded more from the unfolded ensemble than the folded ensemble, and therefore Ŵ
is negative (Figure 2). In the analysis of molecular simulation data, although many quantities,
such as the number of protein-cosolvent hydrogen bonds and changes in the exposed surface
area, provide structural insight into the problem, we stress that calculation of the preferential interaction provides the natural metric for deducing the thermodynamics of the system
considered.
2.5. Emerging View of Urea Denaturation
Recent computational and experimental studies indicate a converging opinion on the role of various
interactions that govern protein denaturation in urea. Canchi and coworkers (56, 57) reported
extensive all-atom replica exchange MD (REMD) simulations to obtain the folding equilibrium
of the Trp-cage miniprotein in the presence of urea, over a wide range of urea concentrations.
Trp-cage, a designed 20-residue protein with a nontrivial fold (78) (Figure 3a), is a system with
Canchi
·
Garcı́a
Annu. Rev. Phys. Chem. 2013.64:273-293. Downloaded from www.annualreviews.org
by RENSSELAER POLYTECHNIC INSTITUTE on 04/05/13. For personal use only.
a wealth of experimental data available (79–82). The folding of Trp-cage can be described by
a two-state model and shows thermodynamic features observed for globular proteins (83). Two
different force fields for the protein—Amber94 (84) and Amber99sb (85)—that differ only in
the backbone torsion angle potential were employed to study the dependence of the computed
driving forces on the unfolded ensemble sampled. The KBFF urea model (58), discussed above,
and the TIP3P water model (86) were used to model the solvent. Figure 3b shows the melting
curves obtained from independent REMD simulations at different urea concentrations. From these
simulations, the authors were able to obtain m-values in good agreement with the experimental
value for Trp-cage (82). Fitting the data globally to a thermodynamic model that accounts for urea
concentration, they showed that denaturation was driven by enthalpy, with a weak dependence of
entropy on urea concentration (56). The two protein models differ significantly in the unfolded
ensemble sampled in the simulations. Figure 3c shows that substantial helical content persists in
the unfolded ensemble obtained in the Amber94 simulations owing to inherent bias in the model.
However, the Amber99sb model shows a two-state separation in helical content between folded
and unfolded ensembles. Sampling the folding equilibrium permits a detailed comparison of the
interaction of urea with both the folded and the unfolded ensembles from which the microscopic
driving forces for the phenomenon can be inferred. Figure 3d shows E = E Unfold − E Fold ,
where E is the interaction energy of the protein with urea molecules in its local domain. Negative
values of E demonstrate that both coulomb and the Lennard-Jones (LJ) interactions of urea
with the protein favor unfolding. Although the coulomb interaction is larger in magnitude than
the LJ interaction, the LJ interaction makes a significant contribution to the difference between
the unfolded and the folded ensembles. Along with the finding that the change in the number of
backbone-urea hydrogen bonds upon unfolding was small, these results imply that denaturation
cannot be explained solely by the hydrogen bonding of urea to the protein.
To address the contribution of backbone and side-chain groups to the free energy of unfolding,
Canchi & Garcı́a (57) calculated the change in the preferential interaction of urea with Trp-cage
upon unfolding. Urea has a larger preferential interaction with the unfolded ensemble than with
the folded ensemble, i.e., Ŵ = Ŵ Unfold − Ŵ Fold > 0, which provides the thermodynamic force
for unfolding. The m-values derived from Ŵ using Equation 5 were in good agreement with
m-values from the LEM analysis, showing thermodynamic consistency in the simulations. The
m-value has been shown to be proportional to the change in exposed surface area upon unfolding
(87), a feature also observed in the above simulations. The change in solvent-accessible surface
area in the Amber94 model is smaller compared with that in Amber99sb owing to higher helical
content, which results in a smaller m-value for Amber94. The simulations also make a thermodynamic argument for the dependence of the unfolded ensemble on urea concentration (70, 71).
Ŵ P was observed to increase with urea concentration, which is necessary as m = RT Ŵ/[C]
is a constant. This is physically feasible only if the unfolded ensemble expands with urea
concentration.
The preferential interaction calculated for the protein (Ŵ P and Ŵ P ) can be decomposed into
backbone and side-chain group contributions, in an additive manner, using the proximity criterion
(88) (Figure 4a). The repeating unit of -C-O-N-Cα - was grouped as the backbone, and the
remaining protein atoms were classified as side chains. The inherent helical bias in the Amber94
model leads to the sampling of conformations in which the backbone is not fully exposed. In
this case, the backbone made no contribution to the increase in the preferential interaction upon
unfolding (Figure 4b). This is a remarkable scenario: It shows that the change in the preferential
interaction of urea and the side chains upon unfolding was sufficient to capture the experimental
thermodynamics (m-value) in the Amber94 model. For the more realistic Amber99sb model in
which the helical bias is eliminated, the side chains, put together, contribute 60% to the m-value
www.annualreviews.org • Cosolvent Effects on Protein Stability
Lennard-Jones (LJ)
potential:
an isotropic 6-12
potential that models
strong repulsion
among molecules at
short distances (Pauli
repulsion) and weak
attraction at larger
distances (van der
Waals interaction)
281
b
Amber94
Amber94
14
Fold
4
Protein
6
2
4
6
Amber99sb
Amber99sb
2
6
8
Fold
Protein
2
Backbone
2
4
[Urea] (M)
4
6
Amber99sb
4
6
2
8
6
Unfold
Side chain
[Urea] (M)
Γ
Γ
4
[Urea] (M)
12
Annu. Rev. Phys. Chem. 2013.64:273-293. Downloaded from www.annualreviews.org
by RENSSELAER POLYTECHNIC INSTITUTE on 04/05/13. For personal use only.
2
[Urea] (M)
6
4
Backbone
2
14
10
Γ
Γ
Γ
Unfold
8
c
8
6
12
10
Amber94
Γ
a
2
[Urea] (M)
4
6
4
2
Side chain
2
4
[Urea] (M)
Figure 4
The preferential interaction of urea with Trp-cage (57). (a) Snapshot of Trp-cage interacting with urea. The protein backbone is
colored in red, whereas the side chains are shown in blue. Urea molecules are shown as spheres, colored according to the group they are
closest to. (b) The preferential interaction in the Amber94 model. The backbone does not make any contribution to the overall change
in preferential interaction upon unfolding. (c) The preferential interaction in the Amber99sb model. The backbone and the side chains
contribute approximately 40% and 60% to the overall change in preferential interaction upon unfolding, respectively. Figure adapted
with permission from Reference 57. Copyright 2011, Elsevier.
for Trp-cage, with the rest accounted for by preferential interactions of the backbone and urea
(Figure 4c). This contrasts with predictions made from transfer model studies, in which the
contribution of side chains has been found to be small or even unfavorable (35, 66).
Recent experimental studies of the interaction of urea with model compounds and amino acids
support the scenario presented above. Lee et al. (37) measured the partial molar volumes and
adiabatic compressibilities of the naturally occurring amino acids in urea solution, as a function of
urea concentration. Using volumetric data in a binding model framework, they demonstrated that
urea has favorable interactions with the backbone as well as most of the side-chain groups. Guinn
et al. (38) took it further by measuring the interaction of urea with a wide range of model compounds, using osmometry to characterize its interaction with various types of molecular surfaces
presented by the protein. Using the solute partitioning model, they reported urea accumulation
in the vicinity of various groups presented by the protein in the following order:
amide O ∼ aromatic C > carboxyl O > amide N > hydroxyl O > aliphatic C.
local
, where m denotes the concentration, is the distribution of
The partition coefficient K p = m
mbulk
urea in local and bulk domains of the solute and ranges from 1.28 for amide O to 1.03 for aliphatic
C. The interaction potentials derived from the above analysis were used to successfully predict the
m-values for a wide range of proteins. The key realization in this work is that even though urea
shows only a weak accumulation around the aliphatic C, this interaction contributes significantly
to the observed effect of urea on protein stability because the composition of the surface exposed in
unfolding is predominantly (∼65%) aliphatic C. The authors estimated a 55% contribution from
the side chains to the m-value for the unfolding of the Trp-cage miniprotein, in good agreement
with the values predicted from simulation studies (57).
282
Canchi
·
Garcı́a
Annu. Rev. Phys. Chem. 2013.64:273-293. Downloaded from www.annualreviews.org
by RENSSELAER POLYTECHNIC INSTITUTE on 04/05/13. For personal use only.
One may ask why the transfer model studies predict a negligible contribution of the side
chains to the free energy of unfolding (35, 66). The transfer free energies (δg tr ) for an amino
acid are calculated from the difference in its solubility between urea and water solutions at the
solubility limit. In the analysis of their data, Auton et al. (66) made nonideality corrections only
for glycine and alanine but not for other amino acids. Guinn et al. (38) have suggested that the
lack of nonideality correction leads to the discrepancy in δg tr reported using solubility data and
osmometry for many amino acids, which can lead to an erroneous contribution from the side
chains to the unfolding free energy. Another possibility lies in the way the δg tr of the side chain is
calculated—by subtracting the δg tr of glycine from the value for the entire amino acid. Comparing
transfer free energies of two solutes directly to determine the δg tr of a functional group does not
account for their differing solubility limits, which can have a significant influence on the measured
value (37).
Lim et al. (89) demonstrated that urea binds to the protein backbone by performing hydrogen
exchange experiments on alanine dipeptide. Although in agreement with all the results described
above, their data cannot be used to attribute the mechanism of urea denaturation to hydrogen
bonding alone as they did not consider the interactions of side chains with urea.
Experimental and computational results now point toward the following mechanism for protein denaturation by urea. Urea is a small organic molecule, which is soluble in water in large
concentrations and incorporates itself well in the hydrogen-bond network of water. It can accept
and donate hydrogen bonds, while also presenting more sites for dispersion interaction than water by the virtue of its molecular structure. Urea accumulates in the vicinity of proteins owing to
favorable direct interaction with them. The direct interaction is a result of the favorable interaction of urea with all protein moieties, including the peptide backbone and most of the side-chain
groups to varying degrees (33, 34, 36–38, 56, 57). The unfolded ensemble has a larger preferential
interaction with urea than the folded ensemble does, which provides an enthalpic driving force
for unfolding. The hydrogen bonding of urea to the protein backbone is an important interaction
in the process of denaturation, but it is not sufficient to entirely account for the destabilization of
proteins in the presence of urea.
3. THE STABILIZATION OF PROTEINS BY TMAO
Protecting osmolytes are small organic molecules that stabilize the folded state of a protein.
TMAO, a widely studied osmolyte, is found in many marine organisms that accumulate urea at
elevated concentrations in their cells to counter osmotic stress (19). The presence of TMAO in
these organisms counteracts the denaturing effect of urea of proteins, typically in a 2:1 concentration ratio of urea and TMAO, and thereby maintains cellular function. The mechanism by
which TMAO stabilizes proteins and counteracts the effect of urea has been investigated, but the
molecular details of these phenomena are still lacking.
Many interesting effects and applications of TMAO have been reported. Quite remarkably,
TMAO can fold proteins that are thermodynamically unfolded to native-like structures with
significant functional activity (90, 91). Based on this observation, investigators used urea-TMAO
mixtures to measure the stability of partially folded proteins and reported that the m-value for urea
denaturation was unaffected by the addition of TMAO (92). Calorimetric studies have indicated
that the enthalpy of unfolding increases linearly with osmolyte concentration (93), suggesting
that a direct interaction mechanism may be involved. The compensation of the opposite effects of
TMAO and urea on the transition temperature, T m , rather than the respective free energies has also
been reported (94). TMAO has been observed to stabilize proteins against pressure denaturation,
which is consistent with its occurrence in deep-sea animals (95). TMAO has also found application
www.annualreviews.org • Cosolvent Effects on Protein Stability
283
Annu. Rev. Phys. Chem. 2013.64:273-293. Downloaded from www.annualreviews.org
by RENSSELAER POLYTECHNIC INSTITUTE on 04/05/13. For personal use only.
in the design of protein-resistant surfaces (96), protein crystallography (97), and the prevention
of misfolding diseases (98).
A wide variety of mechanisms has been proposed to explain the effects of TMAO on protein stability. An indirect mechanism has been suggested in which TMAO stabilizes proteins by enhancing
water structure (99, 100). From dialysis and osmometry experiments, it is known that TMAO is
preferentially excluded from protein surfaces (20, 94). Transfer free energies of amino acids from
water to TMAO solutions have been used to suggest that the observed exclusion results from the
osmophobic interaction of the peptide backbone with TMAO (101). In this model, the compensatory effect on protein stability in urea-TMAO solutions arises from the additivity of the favorable
and unfavorable interactions of urea and TMAO, respectively, with the protein backbone. TMAO
has also been shown to be strongly excluded from hydrophobic surfaces, suggesting that its
exclusion from apolar side chains can contribute significantly to the increased protein stability
(102). An entropic mechanism has been proposed in which TMAO acts as a crowding agent to
favor the more compact native state of proteins through excluded volume effects (103). Neutronscattering experiments have suggested that the compensatory effect of TMAO results from the
direct interaction between TMAO and urea itself without the necessity of TMAO interacting
with the protein directly, although this interaction was later shown to be weak (104, 105).
3.1. TMAO-Water Interactions
Recent experiments have studied the properties of aqueous TMAO solutions. Raman spectroscopy
has indicated that TMAO forms hydrogen bonds with at least three water molecules and has
suggested weak interactions of water with the methyl groups of TMAO (106). Femtosecond midinfrared spectroscopy also suggested long-lived complexes (lifetime >50 ps) of TMAO with two
or three water molecules (100). Using vibrational sum frequency spectroscopy, Sagle et al. (107)
demonstrated that the methyl groups of TMAO orient away from hydrophobic interfaces. This
is a significant finding that implies that not only does the oxide moiety have a tendency to be
hydrated, but so do the methyl groups of TMAO, and this can cause a depletion of cosolvent
molecules near protein surfaces. Koga et al. (108) reached a similar conclusion, showing that
methyl groups need not always be hydophobic, especially when attached to a quaternary nitrogen.
Rösgen & Jackson-Atogi (109) reported activity coefficient data for TMAO and urea-TMAO
solutions, from which solvation properties have been derived using the KB theory. They showed
that TMAO in water behaves effectively as hard spheres, and the binding of urea to TMAO is
nearly random. These studies established that there is a strong interaction of TMAO with water,
which must be considered in explaining the effects of TMAO on protein stability, whereas the
lack of any strong favorable or unfavorable interaction between urea and TMAO may be the basis
for the additivity of urea and TMAO effects.
3.2. Molecular Dynamics Simulation Studies
In contrast to the case of urea, relatively few MD simulation studies have addressed the effects of
TMAO and urea-TMAO mixtures. Simulations of aqueous TMAO show that TMAO remains
hydrated and does not tend to self-aggregate as its concentration is increased (110, 111). The
scenario in which the strengthening of hydrophobic interactions in TMAO solutions leads to
increased protein stability has not been supported by computational investigations. No effect on
hydrophobic interactions was seen in TMAO solutions when compared to water, for small solutes
as well as for the folding of a hydrophobic polymer (112). Disruption of the hydrophobic attraction
between a neopentane pair by the addition of TMAO to both water and urea-water solutions has
284
Canchi
·
Garcı́a
3.3. Simulating the Preferential Exclusion of TMAO from Proteins
Most molecular simulation studies of TMAO in the literature have employed the Kast model for
TMAO (118), developed using ab initio methods. Canchi et al. (119) reported that the Kast model
for TMAO does not show the expected preferential exclusion from protein surfaces and may even
act as a denaturant. Extensive REMD simulations of the Trp-cage miniprotein in TMAO and 2:1
urea-TMAO systems did not show the stabilization of the protein expected due to the addition
of TMAO (Figure 5a). The authors conjectured that this behavior resulted from the inability
of the Kast model to capture the thermodynamics of TMAO solutions. Using the Kast model as
a starting point, they have developed a new model for TMAO by incorporating realistic waterTMAO interactions using osmotic pressure measurements of TMAO solutions, over a range of
a
b
Trp-cage
50
80
4-M urea
4-M urea + 2-M TMAO
40
c
+
30
y
20
=
Kast model for TMAO
10
0
0
100
200
300
Time (ns)
400
Ideal
Experiment
Kast model
Osmotic model
+
z
x
Π (bar)
Number of replicas folded
Annu. Rev. Phys. Chem. 2013.64:273-293. Downloaded from www.annualreviews.org
by RENSSELAER POLYTECHNIC INSTITUTE on 04/05/13. For personal use only.
been reported, implying that TMAO does not protect proteins against denaturation by enhancing
the hydrophobic effect (113, 114). Polyglycine, a model for the peptide backbone, has been shown
to collapse in TMAO solutions, and the exclusion of TMAO from the vicinity of the backbone
was attributed to strong triplet correlations among the solute, TMAO, and water (115). Solvation
free energy calculations of decalanine showed that the unfavorable transfer free energy to TMAO
solutions from water arises from both the van der Waals and electrostatic components, unlike
the case for urea in which the van der Waals component is favorable (116). The only simulation
in the literature of a protein in urea-TMAO solutions was performed by Bennion & Daggett
(117), who attributed the counteracting effect to an indirect mechanism, i.e., the ordering of the
solvent by TMAO. Molecular simulation studies that can capture the effect of TMAO and ureaTMAO mixtures on the protein folding equilibrium, while making rigorous connections with
experimental data (e.g., preferential interaction, m-value), can provide insight into the molecular
underpinnings of these phenomena. Such studies will require the use of physically motivated
force-field parameters for TMAO that address the balance of interactions in a multicomponent
solution, similar to the KBFF model for urea (58). The development of such a model is addressed
in the next section.
60
40
20
500
0
0
0.5
1
1.5
2
2.5
3
[TMAO] (molal)
Figure 5
Model development for TMAO (119). (a) Replica exchange molecular dynamics simulations of Trp-cage in a 2:1 urea/TMAO mixture
showing similar decay to equilibrium as in the simulation with urea alone, indicating that the Kast model is unable to capture the
stabilization of proteins by TMAO. (b) Scheme to calculate osmotic pressure in molecular dynamics simulations. TMAO molecules
(red ) are confined to the central region, while water (blue) can diffuse freely. Osmotic pressure is obtained from the force required to
confine TMAO to the central region. (c) The osmotic model for TMAO obtained by matching the osmotic pressure of TMAO from
simulation to the experimental value, over a range of TMAO concentrations. Figure adapted with permission from Reference 119.
Copyright 2012, American Chemical Society.
www.annualreviews.org • Cosolvent Effects on Protein Stability
285
Annu. Rev. Phys. Chem. 2013.64:273-293. Downloaded from www.annualreviews.org
by RENSSELAER POLYTECHNIC INSTITUTE on 04/05/13. For personal use only.
Mixing rule: specifies
the Lennard-Jones
interaction among
atoms of different
types in an MD
simulation
286
osmolyte concentration. The osmotic pressure, , for an ideal solution is given by the well-known
van’t Hoff law, i.e., Ideal = [C]RT , where [C] is the molar concentration of the solute. Deviation
from ideality is accounted for using the osmotic coefficient, φ, defined as
φ=
Measured
=
.
[C]RT
Ideal
(6)
A positive deviation from ideality (φ > 1) implies solvent-mediated repulsive interactions between
solute molecules, whereas a negative deviation from ideality (φ < 1) indicates a net attractive interaction among solute molecules. The osmotic pressure can be calculated directly from MD
simulations using the scheme shown in Figure 5b (120). An equilibrated solution of a given osmolyte concentration is assembled with systems of solvent (water) of identical size. The resulting
system is simulated by applying a spatial confining potential only on solute particles to restrain
them to occupy their original volume, while solvent molecules are free to diffuse throughout
the system. The confining potential acts as a semipermeable membrane, whereas freely diffusing water mimics the constant solvent chemical potential. The osmotic pressure is then calculated from the average force required to confine the solutes to their original volume during the
simulation.
Figure 5c shows that the measured osmotic pressure for TMAO solutions is larger than the
ideal value (φ > 1), implying a net repulsive interaction between TMAO molecules in solution
(i.e., a strong TMAO-water interaction). The osmotic pressure calculated for TMAO using the
Kast model shows a net repulsive interaction as the TMAO concentration increases, although
not to the extent suggested by experiments at larger concentrations. A new model for TMAO
was obtained that increased the net repulsion among TMAO molecules to match experiments,
termed the osmotic model, by making two changes to the Kast model: (a) scale up the charges to
strengthen the interaction of TMAO with water and (b) weaken the LJ interaction among TMAO
molecules by using a nondefault mixing rule. The osmotic model for TMAO displayed greater
preferential exclusion from protein surfaces than the Kast model for a range of model systems and
proteins, as a consequence of using the osmotic data to capture the thermodynamics of TMAOwater mixtures (Figure 6a). The interaction of TMAO with proteins can be determined using
further experimental inputs, such as the m-value for protein unfolding in TMAO, or by measuring
the change in osmotic pressure resulting from the addition of the protein to the TMAO-water
solution (119).
Furthermore, Canchi et al. (119) examined the role of interaction among cosolvent molecules,
as manifested by the osmotic pressure, in determining its preferential interaction with proteins
(Figure 6b). They studied the osmotic pressure as a function of the polarity of TMAO by scaling
the charges on the reference Kast model. Scaling by a factor α > 1 makes the molecule more
hydrophilic and increases the osmotic pressure of the TMAO-water solution, whereas the opposite
holds true for α < 1. The preferential interaction of TMAO with the protein surface, shown for
Trp-cage in Figure 6b, varies concomitantly with the polarity (or the osmotic pressure) of TMAO.
Making TMAO molecules more attractive leads to an increase in the preferential interaction
with the protein, whereas increasing the solvent-mediated repulsion leads to greater preferential
exclusion from the protein’s vicinity. Qualitatively similar behavior for the preferential interaction
was observed in such simulations with urea.
The intimate connection between preferential exclusion and osmotic pressure provides physical insight into the role of solute-solute interaction in determining the preferential accumulation
or exclusion of cosolvents from proteins. Denaturants, such as urea, have weak favorable interactions with protein surfaces, and a significant concentration of cosolvent is required to denature the proteins. In addition to favorable interactions with proteins, denaturants must also be
Canchi
·
Garcı́a
a
0
0
–5
–5
–2
Γ (r)
Γ (r)
Γ (r)
0
–10
–10
Trp-cage
–4
0
0.5
Kast model
Osmotic model
Ubiquitin
–15
1
0
0.5
r (nm)
Lysozyme
–15
1
0
0.5
r (nm)
c
70
60
50
0.8
1.0
1.2
1.4
Charge scaling, α
1.6
Trp-cage
0.8
2
1.0
0
–2
1.4
0
0.5
r (nm)
1
Osmotic pressure (bar)
0.8
0.9
1.0
1.1
1.2
1.4
Γ (r)
2.5-m TMAO
Π (bar)
Annu. Rev. Phys. Chem. 2013.64:273-293. Downloaded from www.annualreviews.org
by RENSSELAER POLYTECHNIC INSTITUTE on 04/05/13. For personal use only.
b
80
0.6
1
r (nm)
Osmolyte
60
van’t Hoff
Denaturant
45
30
15
0
1
2
3
Concentration (molal)
Figure 6
Osmotic pressure and preferential interaction. (a) The osmotic model for TMAO, showing greater exclusion from protein surfaces than
the Kast model, as a consequence of the osmotic pressure parameterization. (b) The osmotic pressure of the TMAO solution, which can
be controlled in simulations by uniformly scaling the charge on the reference Kast model (α = 1). The preferential interaction of
TMAO with Trp-cage responds concomitantly to the osmotic pressure of the TMAO solution. (c) The osmotic behavior of cosolvents.
It is hypothesized that osmolytes (orange) show a positive deviation from the van’t Hoff law (black), whereas denaturants (red ) show a
negative deviation. Figure adapted with permission from Reference 119. Copyright 2012, American Chemical Society.
able to admit their own excess concentrations in the vicinity (local domain) of the protein, i.e.,
show a negative deviation from ideality with the osmotic coefficient φ ≤ 1. Solvent-mediated
repulsion among osmolyte molecules (φ > 1), apart from the repulsive interaction with protein moieties, adds another penalty that makes the accumulation of osmolytes in the vicinity
of the protein unfavorable. Canchi et al. suggest that denaturants and osmolytes lie on opposite sides of the ideal osmotic pressure curve given by the van’t Hoff law (Figure 6c). From
experimental osmotic pressure data, this scenario is true for urea and TMAO, as well as for
other osmolytes, such as betaine, trehalose, proline, and glycerol (20). By this principle, a mixture of two protecting osmolytes that have repulsive interactions among themselves as reflected
by the nonadditivity of their osmotic pressure may show a synergistic enhancement of protein
stability.
The TMAO model of Canchi et al., obtained by scaling the ab initio charges of the Kast model
(118), represents only one possible parameterization for the molecule. Another approach based
on modifying the LJ parameters of the Kast model also reproduces the osmotic data (R.R. Netz,
personal communication). The robustness of the mechanism of protein stabilization by TMAO
provided by these models is still to be determined.
www.annualreviews.org • Cosolvent Effects on Protein Stability
287
CONCLUSIONS
Annu. Rev. Phys. Chem. 2013.64:273-293. Downloaded from www.annualreviews.org
by RENSSELAER POLYTECHNIC INSTITUTE on 04/05/13. For personal use only.
The control of protein stability by the addition of cosolvents is a biophysical problem of great
fundamental interest, with many practical applications in biotechnology. Understanding the
molecular mechanism of these phenomena has attracted much interest in the literature. The mechanism by which urea denatures proteins has been a contentious issue, but recent computational
and experimental studies point toward an emerging consensus. Urea accumulates in the vicinity
of the protein owing to favorable direct interactions with the protein backbone as well as a range
of amino acid side chains. More favorable protein-urea interactions can be formed in the unfolded
ensemble than in the folded ensemble, which provides an enthalpic driving force for unfolding.
The stabilization of proteins by TMAO and its counteraction of destabilization by urea have
not been satisfactorily explained. Recent studies have shown that the strong interaction of TMAO
with water plays an important role in determining its preferential exclusion from protein surfaces.
Thermodynamic measurements that characterize the interactions of TMAO with amino acids and
model compounds, along with molecular simulations that can capture the effect of TMAO on the
protein folding equilibrium, have the potential to provide insights into the mechanism of these
intriguing phenomena.
DISCLOSURE STATEMENT
The authors are not aware of any affiliations, memberships, funding, or financial holdings that
might be perceived as affecting the objectivity of this review.
ACKNOWLEDGMENTS
This work has been funded by the National Science Foundation (NSF MCB-1050966) and
the National Institutes of Health (GM086801). We thank M.T. Record, G.I. Makhatadze,
D. Thirumalai, D. Rao, and D. Paschek for advice, comments, and suggestions.
LITERATURE CITED
1. Anfinsen CB. 1973. Principles that govern folding of protein chains. Science 181:223–30
2. Bryngelson JD, Onuchic JN, Socci ND, Wolynes PG. 1995. Funnels, pathways, and the energy landscape
of protein folding: a synthesis. Proteins 21:167–95
3. Onuchic JN, Luthey-Schulten Z, Wolynes PG. 1997. Theory of protein folding: the energy landscape
perspective. Annu. Rev. Phys. Chem. 48:545–600
4. Makhatadze GI, Privalov PL. 1995. Energetics of protein structure. Adv. Protein Chem. 47:307–425
5. Privalov PL. 1990. Cold denaturation of proteins. Crit. Rev. Biochem. Mol. Biol. 25:281–305
6. Heremans K. 1982. High-pressure effects on proteins and other biomolecules. Annu. Rev. Biophys. Bioeng.
11:1–21
7. Timasheff SN. 1993. The control of protein stability and association by weak interactions with water:
How do solvents affect these processes? Annu. Rev. Biophys. Biomol. Struct. 22:67–97
8. Kauzmann W. 1987. Protein stabilization: thermodynamics of unfolding. Nature 325:763–64
9. Silva JL, Foguel D, Royer CA. 2001. Pressure provides new insights into protein folding, dynamics and
structure. Trends Biochem. Sci. 26:612–18
10. Royer CA. 2002. Revisiting volume changes in pressure-induced protein unfolding. Biochim. Biophys.
Acta 1595:201–9
11. Hummer G, Garde S, Garcı́a AE, Paulaitis ME, Pratt LR. 1998. The pressure dependence of hydrophobic
interactions is consistent with the observed pressure denaturation of proteins. Proc. Natl. Acad. Sci. USA
95:1552–55
288
Canchi
·
Garcı́a
Annu. Rev. Phys. Chem. 2013.64:273-293. Downloaded from www.annualreviews.org
by RENSSELAER POLYTECHNIC INSTITUTE on 04/05/13. For personal use only.
12. Sarupria S, Ghosh T, Garcı́a AE, Garde S. 2010. Studying pressure denaturation of a protein by molecular
dynamics simulations. Proteins 78:1641–51
13. Roche J, Caro JA, Norberto DR, Barthe P, Roumestand C, et al. 2012. Cavities determine the pressure
unfolding of proteins. Proc. Natl. Acad. Sci. USA 109:6945–50
14. Brandts JF, Oliveira RJ, Westort C. 1970. Thermodynamics of protein denaturation: effect of pressure
on denaturation of ribonuclease A. Biochemistry 9:1038–47
15. Hawley SA. 1971. Reversible pressure-temperature denaturation of chymotrypsinogen. Biochemistry
10:2436–42
16. Paschek D, Gnanakaran S, Garcı́a AE. 2005. Simulations of the pressure and temperature unfolding of
an α-helical peptide. Proc. Natl. Acad. Sci. USA 102:6765–70
17. Paschek D, Hempel S, Garcı́a AE. 2008. Computing the stability diagram of the Trp-cage miniprotein.
Proc. Natl. Acad. Sci. USA 105:17754–59
18. Day R, Paschek D, Garcı́a AE. 2010. Microsecond simulations of the folding/unfolding thermodynamics
of the Trp-cage miniprotein. Proteins 78:1889–99
19. Yancey PH, Clark ME, Hand SC, Bowlus RD, Somero GN. 1982. Living with water stress: evolution
of osmolyte systems. Science 217:1214–22
20. Courtenay ES, Capp MW, Anderson CF, Record MT. 2000. Vapor pressure osmometry studies of
osmolyte-protein interactions: implications for the action of osmoprotectants in vivo and for the interpretation of “osmotic stress” experiments in vitro. Biochemistry 39:4455–71
21. Bolen DW, Rose G. 2008. Structure and energetics of the hydrogen-bonded backbone in protein
folding. Annu. Rev. Biochem. 77:339–62
22. Wang A, Bolen DW. 1997. A naturally occuring protective system in urea-rich cells: mechanism of
osmolyte protection of proteins against urea denaturation. Biochemistry 36:9101–8
23. Kauzmann W. 1959. Some factors in the interpretation of protein denaturation. Adv. Protein Chem.
14:1–63
24. Pace C. 1986. Determination and analysis of urea and guanidine hydrochloride denaturation curves.
Methods Enzymol. 131:266–80
25. Makhatadze GI. 1999. Thermodynamics of protein interactions with urea and guanidinium chloride.
J. Phys. Chem. B 103:4781–85
26. Frank HS, Franks F. 1968. Structural approach to solvent power of water for hydrocarbons: urea as a
structure breaker. J. Chem. Phys. 48:4746–57
27. Dill KA. 1990. Dominant forces in protein folding. Biochemistry 29:7133–55
28. Soper AK, Castner EW, Luzar A. 2003. Impact of urea on water structure: a clue to its properties as a
denaturant? Biophys. Chem. 105:649–66
29. Batchelor JD, Olteanu A, Tripathy A, Pielak GJ. 2004. Impact of protein denaturants and stabilizers on
water structure. J. Am. Chem. Soc. 126:1958–61
30. Rezus YLA, Bakker HJ. 2006. Effect of urea on the structural dynamics of water. Proc. Natl. Acad. Sci.
USA 103:18417–20
31. Kokubo H, Pettitt BM. 2007. Preferential solvation in urea solutions at different concentrations: properties from simulation studies. J. Phys. Chem. B 111:5233–42
32. Stumpe MC, Grubmuller H. 2007. Aqueous urea solutions: structure, energetics, and urea aggregation.
J. Phys. Chem. B 111:6220–28
33. Hua L, Zhou RH, Thirumalai D, Berne BJ. 2008. Urea denaturation by stronger dispersion interactions
with proteins than water implies a two-stage unfolding. Proc. Natl. Acad. Sci. USA 105:16928–33
34. Horinek D, Netz RR. 2011. Can simulations quantitatively predict peptide transfer free energies to urea solutions? Thermodynamic concepts and force field limitations. J. Phys. Chem. A
115:6125–36.
35. Auton M, Bolen DW. 2005. Predicting the energetics of osmolyte-induced protein folding/unfolding.
Proc. Natl. Acad. Sci. USA 102:15065–68
36. Stumpe MC, Grubmuller H. 2007. Interaction of urea with amino acids: implications for urea-induced
protein denaturation. J. Am. Chem. Soc. 129:16126–31
37. Lee S, Shek YL, Chalikian TV. 2010. Urea interactions with protein groups: a volumetric study.
Biopolymers 93:866–79
www.annualreviews.org • Cosolvent Effects on Protein Stability
21. Reviews the
understanding of
cosolvent-dependent
protein stability based
on transfer model
studies.
34. Presents a
thermodynamic analysis
that separates the direct
and indirect effects of
urea in MD simulations.
37. Presents volumetric
measurements of the
interaction of urea with
amino acids.
289
Annu. Rev. Phys. Chem. 2013.64:273-293. Downloaded from www.annualreviews.org
by RENSSELAER POLYTECHNIC INSTITUTE on 04/05/13. For personal use only.
38. Quantifies the
interaction of urea with
various types of
molecular surfaces
presented by a protein
by osmometry.
56. Presents an REMD
simulation of the
folding equilibrium of
the Trp-cage
miniprotein in urea.
57. Addresses the
contribution of
backbone and
side-chain interaction
with urea to the
unfolding free energy of
Trp-cage, based on
preferential interaction
analysis of REMD data.
58. Gives a molecular
model for urea
consistent with the
thermodynamics of
aqueous urea.
290
38. Guinn EJ, Pegram LM, Capp MW, Pollock MN, Record MT. 2011. Quantifying why urea is
a protein denaturant, whereas glycine betaine is a protein stabilizer. Proc. Natl. Acad. Sci. USA
108:16932–37
39. Rossky PJ. 2008. Protein denaturation by urea: slash and bond. Proc. Natl. Acad. Sci. USA 105:16825–26
40. Wallqvist A, Covell DG, Thirumalai D. 1998. Hydrophobic interactions in aqueous urea solutions with
implications for the mechanism of protein denaturation. J. Am. Chem. Soc. 120:427–28
41. Shimizu S, Chan HS. 2002. Origins of protein denatured state compactness and hydrophobic clustering
in aqueous urea: inferences from nonpolar potentials of mean force. Proteins Struct. Funct. Genet. 49:560–
66
42. Oostenbrink C, van Gunsteren WF. 2005. Methane clustering in explicit water: effect of urea on hydrophobic interactions. Phys. Chem. Chem. Phys. 7:53–58
43. Lee ME, van der Vegt NFA. 2006. Does urea denature hydrophobic interactions? J. Am. Chem. Soc.
128:4948–49
44. O’Brien EP, Dima RI, Brooks B, Thirumalai D. 2007. Interactions between hydrophobic and ionic solutes
in aqueous guanidinium chloride and urea solutions: lessons for protein denaturation mechanism. J. Am.
Chem. Soc. 129:7346–53
45. England JL, Pande VS, Haran G. 2008. Chemical denaturants inhibit the onset of dewetting. J. Am.
Chem. Soc. 130:11854–55
46. Zangi R, Zhou RH, Berne BJ. 2009. Urea’s action on hydrophobic interactions. J. Am. Chem. Soc.
131:1535–41
47. Yang LJ, Gao YQ. 2010. Effects of cosolvents on the hydration of carbon nanotubes. J. Am. Chem. Soc.
132:842–48
48. Das P, Zhou RH. 2010. Urea-induced drying of carbon nanotubes suggests existence of a dry globule-like
transient state during chemical denaturation of proteins. J. Phys. Chem. B 114:5427–30
49. Tirado-Rives J, Orozco M, Jorgensen WL. 1997. Molecular dynamics simulations of the unfolding of
barnase in water and 8 m aqueous urea. Biochemistry 36:7313–29
50. Bennion BJ, Daggett V. 2003. The molecular basis for the chemical denaturation of proteins by urea.
Proc. Natl. Acad. Sci. USA 100:5142–47
51. Caballero-Herrera A, Nordstrand K, Berndt KD, Nilsson L. 2005. Effect of urea on peptide conformation
in water: molecular dynamics and experimental characterization. Biophys. J. 89:842–57
52. Stumpe MC, Grubmuller H. 2008. Polar or apolar: the role of polarity for urea-induced protein denaturation. PLoS Comput. Biol. 4:e1000221
53. Smith LJ, Jones RM, van Gunsteren WF. 2005. Characterization of the denaturation of human αlactalbumin in urea by molecular dynamics simulations. Proteins Struct. Funct. Genet. 58:439–49
54. Tobi D, Elber R, Thirumalai D. 2003. The dominant interaction between peptide and urea is electrostatic
in nature: a molecular dynamics simulation study. Biopolymers 68:359–69
55. Klimov DK, Straub JE, Thirumalai D. 2004. Aqueous urea solution destabilizes amyloid β(16–22)
oligomers. Proc. Natl. Acad. Sci. USA 101:14760–65
56. Canchi DR, Paschek D, Garcı́a A. 2010. Equilibrium study of protein denaturation by urea.
J. Am. Chem. Soc. 132:2338–44
57. Canchi DR, Garcı́a AE. 2011. Backbone and side-chain contributions in protein denaturation by
urea. Biophys. J. 100:1526–33
58. Weerasinghe S, Smith PE. 2003. A Kirkwood-Buff derived force field for mixtures of urea and
water. J. Phys. Chem. B 107:3891–98
59. Weerasinghe S, Smith PE. 2003. A Kirkwood-Buff derived force field for sodium chloride in water.
J. Chem. Phys. 119:11342–49
60. Weerasinghe S, Smith PE. 2004. A Kirkwood-Buff derived force field for the simulation of aqueous
guanidinium chloride solutions. J. Chem. Phys. 121:2180–86
61. Weerasinghe S, Smith PE. 2005. A Kirkwood-Buff derived force field for methanol and aqueous methanol
solutions. J. Phys. Chem. B 109:15080–86
62. Kirkwood JG, Buff FP. 1951. The statistical mechanical theory of solutions. 1. J. Chem. Phys. 19:774–77
63. Ben-Naim A. 2006. Molecular Theory of Solutions. New York: Oxford Univ. Press
Canchi
·
Garcı́a
Annu. Rev. Phys. Chem. 2013.64:273-293. Downloaded from www.annualreviews.org
by RENSSELAER POLYTECHNIC INSTITUTE on 04/05/13. For personal use only.
64. Duffy E, Severance D, Jorgenson W. 1993. Urea: potential functions, log p, and free energy of hydration.
Isr. J. Chem. 33:323
65. Tanford C. 2004. Isothermal unfolding of globular proteins in aqueous urea solutions. J. Am. Chem. Soc.
126:1958–61
66. Auton M, Holthauzen LMF, Bolen DW. 2007. Anatomy of energetic changes accompanying ureainduced protein denaturation. Proc. Natl. Acad. Sci. USA 104:15317–22
67. Merchant KA, Best RB, Louis JM, Gopich IV, Eaton WA. 2007. Characterizing the unfolded states
of proteins using single-molecule FRET spectroscopy and molecular simulations. Proc. Natl. Acad. Sci.
USA 104:1528–33
68. O’Brien EP, Ziv G, Haran G, Brooks BR, Thirumalai D. 2008. Effects of denaturants and osmolytes on
proteins are accurately predicted by the molecular transfer model. Proc. Natl. Acad. Sci. USA 105:13403–8
69. Berteotti A, Barducci A, Parrinello M. 2011. Effect of urea on the β-hairpin conformational ensemble
and protein denaturation mechanism. J. Am. Chem. Soc. 133:17200–6
70. Ziv G, Haran G. 2009. Protein folding, protein collapse, and Tanford’s transfer model: lessons
from single-molecule FRET. J. Am. Chem. Soc. 131:2942–47
71. England JL, Haran G. 2011. Role of solvation effects in protein denaturation: from thermodynamics to
single molecules and back. Annu. Rev. Phys. Chem. 62:257–77
72. Yoo TY, Meisburger SP, Hinshaw J, Pollack L, Haran G, et al. 2012. Small-angle X-ray scattering and
single-molecule FRET spectroscopy produce highly divergent views of the low-denaturant unfolded
state. J. Mol. Biol. 418:226–36
73. Timasheff SN. 2002. Protein-solvent preferential interactions, protein hydration, and the modulation
of biochemical reactions by solvent components. Proc. Natl. Acad. Sci. USA 99:9721–26
74. Arakawa T, Timasheff SN. 1982. Preferential interactions of proteins with salts in concentrated solutions.
Biochemistry 21:6545–52
75. Record MT, Anderson CF. 1995. Interpretation of preferential interaction coefficients of nonelectrolytes
and of electrolyte ions in terms of a two-domain model. Biophys. J. 68:786–94
76. Parsegian VA, Rand RP, Rau DC. 2000. Osmotic stress, crowding, preferential hydration and binding:
a comparison of perspectives. Proc. Natl. Acad. Sci. USA 97:3987–92
77. Shukla D, Shinde C, Trout BL. 2009. Molecular computations of preferential interaction coefficients of
proteins. J. Phys. Chem. B 113:12546–54
78. Neidigh JW, Fesinmeyer RM, Andersen NH. 2002. Designing a 20-residue protein. Nat. Struct. Biol.
9:425–30
79. Qiu LL, Pabit SA, Roitberg AE, Hagen SJ. 2002. Smaller and faster: the 20-residue Trp-cage protein
folds in 4 µs. J. Am. Chem. Soc. 124:12952–53
80. Ahmed Z, Beta IA, Mikhonin AV, Asher SA. 2005. UV-resonance Raman thermal unfolding study of
Trp-cage shows that it is not a simple two-state miniprotein. J. Am. Chem. Soc. 127:10943–50
81. Neuweiler H, Doose S, Sauer M. 2005. A microscopic view of miniprotein folding: enhanced folding
efficiency through formation of an intermediate. Proc. Natl. Acad. Sci. USA 102:16650–55
82. Wafer LNR, Streicher WW, Makhatadze GI. 2010. Thermodynamics of the Trp-cage miniprotein
unfolding in urea. Proteins 78:1376–81
83. Streicher WW, Makhatadze GI. 2007. Unfolding thermodynamics of Trp-cage, a 20 residue miniprotein,
studied by differential scanning calorimetry and circular dichroism spectroscopy. Biochemistry 46:2876–80
84. Cornell WD, Cieplak P, Bayly CI, Gould IR, Merz KM Jr, et al. 1995. A second generation force field
for the simulation of proteins, nucleic acids, and organic molecules. J. Am. Chem. Soc. 117:5179–97
85. Hornak V, Abel R, Okur A, Strockbine B, Roitberg A, Simmerling C. 2006. Comparison of multiple
amber force fields and development of improved protein backbone parameters. Proteins 65:712–25
86. Jorgensen WL, Chandrasekhar J, Madura JD, Impey RW, Klein ML. 1983. Comparison of simple
potential functions for simulating liquid water. J. Chem. Phys. 79:926–35
87. Myers JK, Pace CN, Scholtz JM. 1995. Denaturant m-values and heat-capacity changes: relation to
changes in accessible surface areas of protein unfolding. Protein Sci. 4:2138–48
88. Mehrotra PK, Beveridge DL. 1980. Structural analysis of molecular solutions based on quasi-component
distribution functions: application to [H2 CO]aq at 25◦ C. J. Am. Chem. Soc. 102:4287–94
www.annualreviews.org • Cosolvent Effects on Protein Stability
70. Presents a
theoretical analysis that
shows that denaturants
modulate the collapse
transition of proteins.
291
Annu. Rev. Phys. Chem. 2013.64:273-293. Downloaded from www.annualreviews.org
by RENSSELAER POLYTECHNIC INSTITUTE on 04/05/13. For personal use only.
107. Demonstrates a
VSFS study of TMAO
orientation near
hydrophobic surfaces.
109. Presents
thermodynamic data for
TMAO and
urea-TMAO solutions.
112. Examines the
effect of TMAO on
hydrophobic
interactions.
292
89. Lim WK, Rösgen J, Englander SW. 2009. Urea, but not guanidinium, destabilizes proteins by forming
hydrogen bonds to the peptide group. Proc. Natl. Acad. Sci. USA 106:2595–600
90. Baskakov I, Bolen DW. 1998. Forcing thermodynamically unfolded proteins to fold. J. Biol. Chem.
273:4831–34
91. Qu YX, Bolen CL, Bolen DW. 1998. Osmolyte-driven contraction of a random coil protein. Proc. Natl.
Acad. Sci. USA 95:9268–73
92. Mello CC, Barrick D. 2003. Measuring the stability of partly folded proteins using TMAO. Protein Sci.
12:1522–29
93. Attri P, Venkatesu P, Lee MJ. 2010. Influence of osmolytes and denaturants on the structure and enzyme
activity of α-chymotrypsin. J. Phys. Chem. B 114:1471–78
94. Lin TY, Timasheff SN. 1994. Why do some organisms use a urea-methylamine mixture as osmolyte?
Thermodynamic compensation of urea and trimethylamine N-oxide interactions with protein. Biochemistry 33:12695–701
95. Krywka C, Sternemann C, Paulus M, Tolan M, Royer C, Winter R. 2008. Effect of osmolytes on
pressure-induced unfolding of proteins: a high-pressure SAXS study. ChemPhysChem 9:2809–15
96. Anand G, Jamadagni SN, Garde S, Belfort G. 2010. Self-assembly of TMAO at hydrophobic interfaces
and its effect on protein adsorption: insights from experiments and simulations. Langmuir 26:9695–702
97. Mueller-Dieckmann C, Kauffman B, Weiss M. 2011. Trimethylamine N-oxide as a versatile cryoprotective agent in macromolecular crystallography. J. Appl. Crystallogr. 44:433–36
98. Borwankar T, Röthlein C, Zhang G, Techen A, Dosche C, Ignatova Z. 2011. Natural osmolytes remodel
the aggregation pathway of mutant huntingtin exon 1. Biochemistry 50:2048–60
99. Zou Q, Bennion BJ, Daggett V, Murphy KP. 2002. The molecular mechanism of stabilization of proteins
by TMAO and its ability to counteract the effects of urea. J. Am. Chem. Soc. 124:1192–202
100. Hunger J, Tielrooij KJ, Buchner R, Bonn M, Bakker HJ. 2012. Complex formation in aqueous
trimethylamine-N-oxide (TMAO) solutions. J. Phys. Chem. B 116:4783–95
101. Bolen DW, Baskakov IV. 2001. The osmophobic effect: natural selection of a thermodynamic force in
protein folding. J. Mol. Biol. 310:955–63
102. Stanley C, Rau DC. 2008. Assessing the interaction of urea and protein stabilizing osmolytes with the
nonpolar surface of hydroxypropylcellulose. Biochemistry 47:6711–18
103. Cho SS, Reddy G, Straub JE, Thirumalai D. 2011. Entropic stabilization of proteins by TMAO. J. Phys.
Chem. B 115:13401–7
104. Meersman F, Bowron D, Soper AK, Koch MHJ. 2009. Counteraction of urea by trimethylamine N-oxide
is due to direct interaction. Biophys. J. 97:2559–66
105. Meersman F, Bowron D, Soper AK, Koch MHJ. 2011. An X-ray and neutron scattering study of the
equilibrium between trimethylamine N-oxide and urea in aqueous solution. Phys. Chem. Chem. Phys.
13:13765–71
106. Munroe KL, Magers DH, Hammer NI. 2011. Raman spectroscopic signatures of noncovalent interactions between trimethylamine N-oxide (TMAO) and water. J. Phys. Chem. B 115:7699–707
107. Sagle LB, Cimatu K, Litosh VA, Liu Y, Flores SC, et al. 2011. Methyl groups of trimethylamine
N-oxide orient away from hydrophobic interfaces. J. Am. Chem. Soc. 133:18707–12
108. Koga Y, Westh P, Nishikawa K, Subramanian S. 2011. Is a methyl group always hydrophobic? Hydrophilicity of trimethylamine-N-oxide, tetramethyl urea and tetramethylammonium ion. J. Phys. Chem.
B 115:2995–3002
109. Rösgen J, Jackson-Atogi R. 2012. Volume exclusion and H-bonding dominate the thermodynamics and solvation of trimethylamine-N-oxide in aqueous urea. J. Am. Chem. Soc. 134:3590–97
110. Fornili A, Civera M, Sironi M, Fornili SL. 2003. Molecular dynamics simulation of aqueous solutions
of trimethylamine-N-oxide and tert-butyl alcohol. Phys. Chem. Chem. Phys. 5:4905–10
111. Sinibaldi R, Casieri C, Melchionna S, Onori G, Segre AL, et al. 2006. The role of water coordination in
binary mixtures: a study of two model amphiphilic molecules in aqueous solutions by molecular dynamics
and NMR. J. Phys. Chem. B 110:8885–92
112. Athawale MV, Dordick JS, Garde S. 2005. Osmolyte trimethylamine-N-oxide does not affect the
strength of hydrophobic interactions: origin of osmolyte compatibility. Biophys. J. 89:858–66
Canchi
·
Garcı́a
Annu. Rev. Phys. Chem. 2013.64:273-293. Downloaded from www.annualreviews.org
by RENSSELAER POLYTECHNIC INSTITUTE on 04/05/13. For personal use only.
113. Paul S, Patey GN. 2007. The influence of urea and trimethylamine-N-oxide on hydrophobic interactions.
J. Phys. Chem. B 111:7932–33
114. Paul S, Patey GN. 2008. Hydrophobic interactions in urea: trimethylamine-N-oxide solutions. J. Phys.
Chem. B 112:11106–11
115. Hu CY, Lynch GC, Kokubo H, Pettitt BM. 2010. Trimethylamine N-oxide influence on the backbone
of proteins: an oligoglycine model. Proteins 78:695–704
116. Kokubo H, Hu CY, Pettitt BM. 2011. Peptide conformational preferences in osmolyte solutions: transfer
free energies of decaalanine. J. Am. Chem. Soc. 133:1849–58
117. Bennion BJ, Daggett V. 2004. Counteraction of urea-induced protein denaturation by trimethylamine
N-oxide: a chemical chaperone at atomic resolution. Proc. Natl. Acad. Sci. USA 101:6433–38
118. Kast KM, Brickmann J, Kast SM, Berry RS. 2003. Binary phases of aliphatic N-oxides and water: force
field development and molecular dynamics simulation. J. Phys. Chem. A 107:5342–51
119. Canchi DR, Jayasimha P, Rau DC, Makhatadze GI, Garcia AE. 2012. Molecular mechanism for
the preferential exclusion of TMAO from protein surfaces. J. Phys. Chem. B 116:12095–104
120. Luo Y, Roux B. 2010. Simulation of osmotic pressure in concentrated aqueous salt solutions. J. Phys.
Chem. Lett. 1:183–89
www.annualreviews.org • Cosolvent Effects on Protein Stability
119. Proposes a
molecular model for
TMAO based on
osmotic data.
293
Contents
Annual Review of
Physical Chemistry
Volume 64, 2013
Annu. Rev. Phys. Chem. 2013.64:273-293. Downloaded from www.annualreviews.org
by RENSSELAER POLYTECHNIC INSTITUTE on 04/05/13. For personal use only.
The Hydrogen Games and Other Adventures in Chemistry
Richard N. Zare ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 1
Once upon Anion: A Tale of Photodetachment
W. Carl Lineberger ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣21
Small-Angle X-Ray Scattering on Biological Macromolecules
and Nanocomposites in Solution
Clement E. Blanchet and Dmitri I. Svergun ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣37
Fluctuations and Relaxation Dynamics of Liquid Water Revealed
by Linear and Nonlinear Spectroscopy
Takuma Yagasaki and Shinji Saito ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣55
Biomolecular Imaging with Coherent Nonlinear
Vibrational Microscopy
Chao-Yu Chung, John Boik, and Eric O. Potma ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣77
Multidimensional Attosecond Resonant X-Ray Spectroscopy
of Molecules: Lessons from the Optical Regime
Shaul Mukamel, Daniel Healion, Yu Zhang, and Jason D. Biggs ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 101
Phase-Sensitive Sum-Frequency Spectroscopy
Y.R. Shen ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 129
Molecular Recognition and Ligand Association
Riccardo Baron and J. Andrew McCammon ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 151
Heterogeneity in Single-Molecule Observables in the Study
of Supercooled Liquids
Laura J. Kaufman ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 177
Biofuels Combustion
Charles K. Westbrook ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 201
Charge Transport at the Metal-Organic Interface
Shaowei Chen, Zhenhuan Zhao, and Hong Liu ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 221
Ultrafast Photochemistry in Liquids
Arnulf Rosspeintner, Bernhard Lang, and Eric Vauthey ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 247
v
Cosolvent Effects on Protein Stability
Deepak R. Canchi and Angel E. Garcı́a ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 273
Discovering Mountain Passes via Torchlight: Methods for the
Definition of Reaction Coordinates and Pathways in Complex
Macromolecular Reactions
Mary A. Rohrdanz, Wenwei Zheng, and Cecilia Clementi ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 295
Water Interfaces, Solvation, and Spectroscopy
Phillip L. Geissler ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 317
Annu. Rev. Phys. Chem. 2013.64:273-293. Downloaded from www.annualreviews.org
by RENSSELAER POLYTECHNIC INSTITUTE on 04/05/13. For personal use only.
Simulation and Theory of Ions at Atmospherically Relevant Aqueous
Liquid-Air Interfaces
Douglas J. Tobias, Abraham C. Stern, Marcel D. Baer, Yan Levin,
and Christopher J. Mundy ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 339
Recent Advances in Singlet Fission
Millicent B. Smith and Josef Michl ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 361
Ring-Polymer Molecular Dynamics: Quantum Effects in Chemical
Dynamics from Classical Trajectories in an Extended Phase Space
Scott Habershon, David E. Manolopoulos, Thomas E. Markland,
and Thomas F. Miller III ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 387
Molecular Imaging Using X-Ray Free-Electron Lasers
Anton Barty, Jochen Küpper, and Henry N. Chapman ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 415
Shedding New Light on Retinal Protein Photochemistry
Amir Wand, Itay Gdor, Jingyi Zhu, Mordechai Sheves, and Sanford Ruhman ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 437
Single-Molecule Fluorescence Imaging in Living Cells
Tie Xia, Nan Li, and Xiaohong Fang ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 459
Chemical Aspects of the Extractive Methods of Ambient Ionization
Mass Spectrometry
Abraham K. Badu-Tawiah, Livia S. Eberlin, Zheng Ouyang,
and R. Graham Cooks ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 481
Dynamic Nuclear Polarization Methods in Solids and Solutions to
Explore Membrane Proteins and Membrane Systems
Chi-Yuan Cheng and Songi Han ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 507
Hydrated Interfacial Ions and Electrons
Bernd Abel ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 533
Accurate First Principles Model Potentials for
Intermolecular Interactions
Mark S. Gordon, Quentin A. Smith, Peng Xu,
and Lyudmila V. Slipchenko ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 553
vi
Contents
Structure and Dynamics of Interfacial Water Studied by
Heterodyne-Detected Vibrational Sum-Frequency Generation
Satoshi Nihonyanagi, Jahur A. Mondal, Shoichi Yamaguchi,
and Tahei Tahara ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 579
Molecular Switches and Motors on Surfaces
Bala Krishna Pathem, Shelley A. Claridge, Yue Bing Zheng,
and Paul S. Weiss ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 605
Annu. Rev. Phys. Chem. 2013.64:273-293. Downloaded from www.annualreviews.org
by RENSSELAER POLYTECHNIC INSTITUTE on 04/05/13. For personal use only.
Peptide-Polymer Conjugates: From Fundamental Science
to Application
Jessica Y. Shu, Brian Panganiban, and Ting Xu ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 631
Indexes
Cumulative Index of Contributing Authors, Volumes 60–64 ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 659
Cumulative Index of Article Titles, Volumes 60–64 ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 662
Errata
An online log of corrections to Annual Review of Physical Chemistry articles may be
found at http://physchem.annualreviews.org/errata.shtml
Contents
vii