PCCP

View Article Online

PAPER View Journal | View Issue

Ordered hydration layer mediated ice

adsorption of a globular antifreeze protein:
Published on 14 August 2019. Downloaded by University of Perugia on 5/22/2023 8:46:37 AM.

Cite this: Phys. Chem. Chem. Phys.,

2019, 21, 19298 mechanistic insight
Sandipan Chakraborty and Biman Jana *

Received 3rd June 2019,
Accepted 14th August 2019
DOI: 10.1039/c9cp03135a
rsc.li/pccp
The ice/water interface recognition mechanism of antifreeze proteins (AFPs) is highly contentious.
Conventionally, protein adsorption on a solid surface is primarily driven by the polar interactions
between the hydrophilic residues of the protein and interfacial water of the solid surface. Ice surface
recognition by a type III AFP is surprising in this context where the ice binding surface (IBS) is
hydrophobic. The present study provides molecular insight into the unusual interface recognition
phenomenon of a type III AFP (QAE isoform) from Macrozoarces americanus. Potential of mean force
calculations show that the type III AFP adsorbs on the ice surface mediated through a layer of ordered
water. Molecular dynamics simulations at lower than ambient temperature reveal that the flat
hydrophobic IBS induces ordering of water. The excellent geometrical synergy between the hydration
water structure around the IBS and water arrangements on the pyramidal surface favours adsorption on
the pyramidal plane. Mutations that interrupt the hydration shell water ordering essentially lead to less
efficient adsorption, which greatly reduces the anti-freezing activity of the AFP. Binding free energy
calculations of the wild-type and several mutant AFPs reveal that the binding affinity is linearly correlated
with the experimentally observed thermal hysteresis activity. Therefore, binding to a specific ice plane
with considerable affinity is the dictating factor of the anti-freeze activity for a type III AFP. Mechanistic
insights into the ice binding process of the wild-type and different mutant AFPs obtained from this study
pave the way for rational design of type III variants with much improved activity, which possesses ample
industrial applicability, particularly in cryo-preservation.

1. Introduction
Understanding the mechanism of protein and peptide adsorption
on solid/liquid interfaces is a subject of recent interest due
to its immense applications in materials science, medicine,
and technology.1–4 Surface adsorption plays a pivotal role
in material design and application, bio-mineralization and
biosensor design.3,5–8 The structure and activity of proteins,
particularly enzymes, on solid surfaces have been studied
extensively to understand the effect of surface adsorption on
the biological function of the proteins.6,9–12 Notably the fate of
nanoparticles in vivo depends strongly on their interactions
with different body proteins. Thus understanding the mechanism
of different interface recognition by proteins emerges as a topic of
intense research interest. Although efforts were started in the
1960s by Vroman et al.,13 exploration of molecular understanding
of the process of adsorption has gained tremendous attention in
School of Chemical Sciences, Indian Association for the Cultivation of Science,
Jadavpur, Kolkata-700032, India. E-mail: pcbj@iacs.res.in; Fax: +91 33 2473 2805;
Tel: +91 33 2473 4971
the last two decades.14–18 Several models have been proposed.16–19
In earlier work of protein adsorption on a polyethylene oxide
surface, Jeon et al. proposed diffusion mediated protein adsorp-
tion where van der Waals attraction plays an important role.17
Later, Szleifer et al. proposed a theoretical framework for protein
adsorption on surfaces based on generalized single-chain mean
field theory where the surface coverage, the concentration of the
bulk protein and the nature of the solvent dictate the adsorption
isotherm.18 Norde et al. on the other hand highlighted that the
balance between conformational entropy and specific interactions
governs protein adsorption on surfaces.20 All these models explain
the surface adsorption at the macroscopic level and the process of
surface recognition of proteins at the molecular level is still poorly
understood. Recent work of Penna et al.14 and Yu et al.15 put
forward a three-step model where in the first step there is a biased
diffusion of the protein towards the interface, which is followed by
its anchoring in the second hydration shell of the surface using
the hydrophilic surface residues and then popping into the first
hydration shell of the solid surface. Finally, the protein rearranges
itself to lock down on the solid surface by optimizing the protein–
solid surface interactions. The roles of polar side chains are

19298 | Phys. Chem. Chem. Phys., 2019, 21, 19298--19310 This journal is © the Owner Societies 2019

Paper
crucial in this three-step model of protein adsorption. However,
type III antifreeze proteins (AFPs) are a surprise in this context.
This class of AFP uses a hydrophobic surface to recognize the
ice/water interface, which is essentially polar in nature.21
AFPs are evolutionarily selected for ice/water interface recogni-
tion, which is required for the survival of living organisms in
the cold Antarctic winter. These bio-macromolecules have been
isolated from Antarctic fish, insects, bacteria, yeast and even
plants.22–31 AFPs are commonly classified as fish type I, II, III,
and IV, insect AFPs and antifreeze glycoproteins (AFGPs).
Published on 14 August 2019. Downloaded by University of Perugia on 5/22/2023 8:46:37 AM.
Different classes of AFPs bear no structural similarities and
they are evolutionary independent from each other.32
The mechanism of ice recognition by AFPs is highly
controversial. The ice binding surface (IBS) of type I and insect
AFPs contains an array of polar residues. Their repeat distances
match the oxygen atom repeat distances of the particular ice
plane where they specifically bind.33,34 This observation is the
origin of the first proposed mechanism, known as the ‘‘hydrogen
bond matching hypothesis’’.33 However, the hypothesis was later
rejected when mutational data on type I AFPs were available.
Substitution of equidistant Thr residues at the IBS with Ser
(similar hydrogen bonding ability) leads to complete loss of
activity; however the AFP still remains B30% active upon
replacement of Thr with Val, which does not possess hydrogen
bonding side chains.22 Later, a surface complementarity matching
hypothesis has been proposed motivated by the observation
that many AFPs possess a flat IBS and substitution of the IBS
residues by amino acids with bulkier side chains diminishes
the activity.21,35,36 However, this mechanism is also highly
criticized since the ice/water interface is not a rigid surface to
provide enough strength for AFP adsorption, which is known to
be irreversible in nature.37
Recently, AFPs have come into prominence due to their
unusual ability to control hydration water dynamics, particularly
around the IBS.38–41 Initially, it has been argued that an AFP
induces long range perturbations to water dynamics which alter
the local ice growth kinetics at the ice/water interface40,42 and
even induce ice melting.43,44 However, microfluidic experiments
and fluorescence microscopy clearly showed that the AFPs
adhere to the ice surface10,37,45–47 and the binding is nearly
irreversible in nature.45 Later it has been demonstrated that
both short-range interactions and long-range protein induced
water dynamics dictate the antifreeze activity, at least for the
insect AFPs.41 However, atomistic insight into the process of
adsorption is largely lacking. Very recently, we have proposed a
clathrate mediated ice adsorption mechanism for type I48 and
insect AFPs.49–53 According to the mechanism, the IBS first
adheres to the ice/water interface mediated through the ordered
hydration layer and there is formation of ordered water cages
around the IBS. In insect AFPs, the presence of threonine
repeats at the IBS stabilizes a part of the water cages, known
as ‘‘anchored clathrate water’’ (ACW),54 which forms dual
hydrogen bonds involving both the AFP and ice surface, thus
anchoring the AFP on the ice surface.
Type III AFPs belong to the category of globular proteins.
Unlike the type I and insect AFPs, type III AFPs do not have any
This journal is © the Owner Societies 2019
View Article Online
PCCP
threonine repeats. Moreover, the identified IBS is hydrophobic
in nature.21,55,56 The binding plane specificity of type III AFPs is
another controversial area of research. Initially, prism plane
binding was proposed, based on ice etching experiments.55,57
Later, Antson et al. reported modified ice etching data for type
III AFPs which indicated its binding to different ice planes that
are parallel or inclined at a small angle to the c-axis.58 Thus type
III AFPs are thought to be a non-basal multiple plane binder.
A previous simulation study revealed slow water dynamics
around the IBS of type III AFP59 and even the presence of
ice-like ordered hydration water within the first hydration shell
was also reported.60,61 The presence of such an ordered water
population was also evident from vibrational sum frequency
generation spectroscopy.62 Grabowska et al.63 and Xu et al.64
hinted that both long range solvent perturbation and short
range local interactions determine the hysteresis activity.
Mahatabuddin et al. recently reported the crystal structure of
an improved activity variant of a type III AFP from notched-fin
eelpout (A20I mutant) where a polypentagonal ice-like water
network is clearly evident on the IBS of the AFP, which binds to
the pyramidal binding plane.65 However, Sun et al. using fusion
protein crystallography showed that the water layer around the
IBS of a type III AFP which binds to the pyramidal plane is
sparse and does not form a robust network.66 Notably, crystal
water molecules are strongly influenced by crystal packing
artifacts and the conditions of the experiments (temperature,
buffer, salt concentrations). Also, a pentagonal ice-like water
structure has been reported in many other non-antifreeze proteins
too, like crambin,67 bacterial malate dehydrogenase,68 etc. There-
fore, the prospects of hydration water mediated ice recognition for
type III AFPs remain controversial, particularly due to the fact that
the structure of type III AFPs bound to an ice surface in the
presence of an ice/water interface is yet to be characterized. Thus
molecular-scale insight into the process of adsorption of a type III
AFP on an ice surface still remains elusive.
The present work for the first time provides the atomistic
details of the ice binding mechanism of a type III AFP on
different ice planes using equilibrium simulations and free
energy calculations. The structure and thermodynamic aspects
of type III AFP adsorption on the ice surface have been critically
explored. Our objectives are to explore in critical detail how the
hydrophobic protein surface adsorbs on the ice/water interface
and the role of hydrophobic hydration in the process of
adsorption. We have further applied our proposed adsorption
model to explain the experimentally observed activity data of
several type III AFP mutants.
2. Methods
All the molecular dynamics simulations and free energy calcula-
tions were performed using the GROMACS package69–71 in
combination with the OPLS/AA72–74 force field and SPC/E water
model.75 Validations of the water model and force field in low
temperature simulations of the AFPs were discussed in detail in
our recent publications.48,49 Previously, Bryk et al. demonstrated
Phys. Chem. Chem. Phys., 2019, 21, 19298--19310 | 19299

PCCP
that the basal and prism plane of ice interfaces can be appropriately
mimicked by SPC/E water at a temperature of 225  5 K. Within
this range, the density, translational, orientational and dynamic
order parameters smoothly and continuously change from the
crystal to the liquid across the interface.76 All the visualizations
were carried out using the Visual Molecular Dynamics (VMD)
package.77
2.1. Preparation of the systems
The structure of the type III AFP from Macrozoarces americanus
Published on 14 August 2019. Downloaded by University of Perugia on 5/22/2023 8:46:37 AM.
was obtained from the protein data bank (PDB ID: 1MSI).55
Residue protonation states were assigned according to neutral
pH. All the heteroatoms and crystallographic water molecules
were not considered in the simulation. Different AFP mutants
(A16M, L19A//V41A, N14S, and T18N) were constructed by
mutating a selected residue of the wild-type protein to the
required residue at the particular position. An appropriate side
chain rotamer was selected from the rotamer library where
minimum steric clashes were observed.
2.2. Equilibrium simulations of wild-type and mutant type III
AFPs
Wild-type and different mutant AFPs were energy minimized
in vacuo using the steepest descent algorithm. Each system was
then immersed in a cubic box containing SPC/E water in such a
way that the minimum distance between the protein atoms and
box edges was greater than 10 Å. Minimizations in aqueous
environments were then performed for all the systems using
another 500 steps of energy minimization using the steepest
descent algorithm. Then a 1 ns position restraint simulation
was performed for each AFP in the isothermal–isobaric (NPT)
ensemble. Protein backbone atoms were restrained using a
harmonic potential with a force constant of 1000 kJ mol1 nm2
in all three dimensions, however, water molecules were allowed to
move freely. All the simulations were performed under periodic
boundary conditions.
To study the conformational dynamics of the wild-type and
mutant AFPs, we performed equilibrium simulations at 298 K.
The temperature was kept constant by coupling to a thermostat
using the V-rescale algorithm78 with a time constant for
coupling set to 0.1 ps. The pressure was kept constant at 1 bar
by employing the isotropic Parrinello–Rahman barostat with a
time-constant of 2 ps. Electrostatic interactions were calculated
using the particle mesh Ewald summation (PME) method79 with
default values for the grid spacing and a 10 Å distance cut-off
was used to cut short-range electrostatic interactions. Finally, a
500 ns production simulation was carried out for each system
in the NPT ensemble using the same parameters and the
coordinates of the simulated systems were saved every 20 ps.
For hydration study, the energy minimized solvated struc-
ture of wild-type AFP was subjected to 1 ns position restrained
equilibrations at four different temperatures (210 K, 225 K,
250 K, and 298 K) in the NPT ensemble. The same parameters
for temperature and pressure coupling were used. Final 50 ns
production runs were carried out in the NPT ensemble at
four different temperatures. During the production run, the
19300 | Phys. Chem. Chem. Phys., 2019, 21, 19298--19310
View Article Online
Paper
translational movement of the protein was restricted. The
trajectory was saved at every 1 ps. In the case of the four
different mutants, similar protocols were followed during
the 50 ns simulation for each system in the NPT ensemble.
All the simulations for the mutant AFPs were performed only at
a single temperature (225 K) at which the PMF calculations
were carried out.
All the analyses were performed using the available analysis
tools in GROMACS. For RMSD calculations, structural alignment
was carried out on Ca atoms and the RMSD was calculated on all
heavy atoms of the protein from the simulation. The RMSF was
calculated on all heavy atoms of the protein from the simulation.
2.3. Molecular docking of AFPs with ice planes
The energy minimized wild-type AFP was initially docked to
three different ice planes (basal, prism and pyramidal planes)
using a rigid body docking procedure implemented in the
PatchDock webserver.80 Docking generates many possible
docked AFP–ice complexes. The possible docked complex where
the experimentally known binding sites oriented towards the
ice surface was chosen. In the case of mutant AFPs, adsorption
was considered only on the pyramidal plane. The wild-type
AFP–pyramidal ice complex obtained from molecular docking
was selectively mutated at the particular IBS residue to generate
mutant AFP–pyramidal ice complexes with the aid of mutagenesis
tools implemented in the VMD package.77
2.4. Calculations of potential of mean force (PMF) for
wild-type and mutant AFP adsorption on ice planes
For the wild-type AFP, the free energy of binding was calculated
for basal, prism and pyramidal plane adsorption. While for the
mutant systems, the free energy of binding was calculated for
the pyramidal plane only. AFP–ice docked complexes were
considered initially. Each docked complex was prepared and
equilibrated. Briefly, each system was aligned in such a way
that the ice slab was on the X–Y plane and the width of the slab
was along the Z-axis. Then the system was solvated in a
rectangular box containing SPC/E water. The box dimension
was chosen in such a way that in the X–Y plane it did not collide
with its image and along the Z-axis the box distance was greater
than double the final pull distance. For wild type AFP adsorp-
tion on basal and prism planes the box dimensions were
80  80  100 Å3 and for wild-type and mutant AFP adsorption
on the pyramidal plane the box dimensions were chosen as
100  80  100 Å3. All the simulations were performed under
periodic boundary conditions. All the systems were initially
energy minimized using the steepest descent algorithm followed
by 1 ns position restraint dynamics in the NVT ensemble at
225 K. During the simulation the ice slab was frozen. The
temperature was kept constant by coupling to a thermostat
using the V-rescale algorithm with a time constant for coupling
set to 0.1 ps. Electrostatic interactions were calculated using the
PME method. Then finally a 3 ns NVT simulation was carried out
for each system at 225 K where both the proteins and water
molecules were allowed to move freely, while the ice slab was
kept frozen and the simulation parameters remain unaltered.
This journal is © the Owner Societies 2019

Paper
The umbrella sampling technique was applied to construct
the potential of mean force (PMF) curve for wild-type and
mutant AFP adsorption to the ice surface. The AFP and ice slab
center of mass distance along the Z-direction was considered as
the reaction coordinate. The AFP was then pulled from the
respective ice plane along the Z-direction using the umbrella
sampling technique. The interval between two successive
windows was 1 Å along the reaction coordinate. In each
window, a force constant of 2000 kJ mol1 nm2 was applied
to maintain the required reaction coordinate value using an
Published on 14 August 2019. Downloaded by University of Perugia on 5/22/2023 8:46:37 AM.
umbrella potential. In each umbrella window, a 1 ns equili-
bration simulation and a 5 ns production run were performed
in the NVT ensemble. All the simulation parameters remained
the same as mentioned in the equilibration process. Convergence
of the simulation in each window was confirmed by performing
block averaging. The weighted histogram analysis method
(g_wham) was then used to construct the PMF profiles.81
Sufficient overlaps among the distributions of reaction coordinates
were confirmed by histogram analysis.
2.5. Equilibrium simulation of wild-type and mutant AFP–ice
complexes
Structures of the AFP–ice complexes corresponding to the PMF
minima were further simulated in the NVT ensemble at 225 K
for 30 ns to characterize the adsorbed state. All the simulation
parameters remained unaltered as mentioned above. For all the
simulations the ice slab was kept frozen in all three dimensions.
3. Results and discussion
In this work, we have studied the molecular mechanism of ice
recognition of a type III AFP (QAE isoform) from Macrozoarces
americanus. We have considered the ice binding surface (IBS)
based on previously reported experimental mutation data and
initial docking studies.21,55 Garnham et al. also showed that
there are two adjacent ice recognition sites aligned with an
angle B1501.82 Here, we have considered all the residues of
both recognition sites to define the IBS. Residues 9, 12–16, 18–
20, 41, 44 and 51 are considered to be involved in the IBS.21,55,82
Other residues of the protein are considered as the non-IBS
plane. A pictorial representation of the IBS and non-IBS is
shown in Fig. 1A. Unlike other classes of AFP, the IBS of type
III AFPs is relatively flat and electrostatic surface potential
calculations reveal that the IBS is hydrophobic in comparison
to the non-IBS plane (Fig. 1B). There is a patch of weak
negatively charged surface at the tip of the IBS indicated by
the circle in the figure. This is primarily contributed by Asn14,
Thr15, Thr18, and Gln44.
3.1. Type III AFP has the ability to induce water ordering
around the IBS
We have investigated the water ordering around both the IBS
and non-IBS and its variation upon lowering the temperature.
AFPs are active around the freezing temperature of water.
We have used the SPC/E water model to mimic the hydration
This journal is © the Owner Societies 2019
View Article Online
PCCP
Fig. 1 Structure of the type III AFP from Macrozoarces americanus shown
as a cyan cartoon (A). Regions of the IBS are shown in a blue transparent
surface representation. The rest of the protein is considered as the non-
IBS. (B) The electrostatic surface potential of the type III AFP is shown.
Regions of positive and negative potentials are shown in blue and red
colours, respectively. The IBS with sparse negative electrostatic potential is
shown within the circle.
around the AFP. The melting point of the SPC/E water model is
B215 K. Therefore, we have simulated the AFP at four different
temperatures (210 K, 225 K, 250 K, and 298 K) and analysed the
water ordering within the first hydration shell (Fig. 2).
As is evident from the figure, there is a preferential accumu-
lation of water around the IBS compared to the non-IBS plane at
all temperatures. The peak appears at B3.7 Å, which is larger
than the hydrogen bond donor–acceptor distance limit (3 Å),
which reinforces the fact that these water molecules are not
hydrogen bonded to the IBS planes. We have further analysed
the ordering of these water molecules using the O–O–O angle
distribution. This order parameter is a highly sensitive one that
can differentiate between interstitial and tetrahedrally oriented
water.83,84 The distribution is characterized by two peaks. The
low angle peak B601 is related to the interstitial water and the
peak at B100–1041 signifies tetrahedrally ordered hydrogen
bonded water. The most common arrangement of hydrogen
bonded water molecules around an oxygen atom of a central
water is nearly tetrahedral. Due to the dynamic nature of liquid
water, hydrogen bonds vary in distance as well as in symmetry,
which resulted in the formation of transient irregular tetrahedral
arrangements. A representative tetrahedral arrangement of water
is shown in the inset of Fig. 2. However, in the case of ice, the
perfect ordering of water molecules in the crystal lattice makes
the hydrogen bonds highly symmetrical, which results in the
formation of a perfect tetrahedron. Room temperature liquid
water exhibits the peak of the O–O–O angle distribution of
tetrahedral water at B991. A shift of the peak maximum towards
higher angle signifies more tetrahedral ordering and an increase
in magnitude signifies an increase in the population of such
ordered water. We have noticed that the higher angle peak
maximum of the O–O–O angle distribution of the first hydration
shell water around the IBS progressively shifts toward a higher
value and its magnitude increases with a decrease in temperature
Phys. Chem. Chem. Phys., 2019, 21, 19298--19310 | 19301
 View Article Online

PCCP Paper
Published on 14 August 2019. Downloaded by University of Perugia on 5/22/2023 8:46:37 AM.

Fig. 2 Analysis of hydration water at four different temperatures around the IBS and non-IBS (nIBS) planes of the type III AFP. IBS and non-IBS planes are
represented in black and red colours respectively. Radial distributions of water (oxygen) molecules around different planes of the AFP at four different
temperatures are shown and the corresponding O–O–O angle distributions among water molecules within the first hydration shell (within 5.5 Å) around
each surface of the protein are also shown in the inset. A representative tetrahedral arrangement of water is shown in the inset.

compared to the non-IBS plane. This trend clearly suggests that

upon lowering the temperature water molecules around the IBS
adopt a more ordered structure. The population of tetrahedrally
ordered water is higher around the IBS compared to the non-IBS
plane at all the temperatures, which is particularly pronounced in
low temperature regions. Noticeably, the higher angle peak is
shifted by 21 in the case of IBS compared to the non-IBS plane,
which is more pronounced at low temperatures and is related
to the hydrophobic hydration of the IBS in the low temperature
region.
The water structure around the IBS obtained at 30 ns of
equilibrium simulation at 210 K is shown in Fig. 3. The water
molecules within the first hydration shell of the IBS are highly
ordered and there is the formation of water cages stabilized
through a highly cooperative hydrogen bond network (Fig. 3).
Notably, Meister et al. also noticed water ordering in the
aqueous solution of a type III AFP, even above the freezing
temperature, by using femtosecond sum frequency generation
spectroscopy.62

3.2. Type III AFP preferentially adsorbs on the pyramidal
plane mediated through a layer of water
We have performed potential of mean force calculations using
umbrella sampling techniques to delineate the binding plane
Fig. 3 Structure of the water shell around the IBS of the type III AFP.
The IBS is shown as a blue transparent surface and the protein is rendered
as a cyan cartoon. Oxygen atoms of the hydration layer water are rendered
in a red vdW mode and the hydrogen bond between them is shown as a
red line.

19302 | Phys. Chem. Chem. Phys., 2019, 21, 19298--19310 This journal is © the Owner Societies 2019
 View Article Online

Paper PCCP

specificities of this class of AFP. We have considered three
different ice planes in this study, namely basal, prism and
pyramidal planes.
Fig. 4A reveals that the type III AFP preferentially binds to
the pyramidal plane and the order of binding affinities towards
different ice planes is pyramidal plane 4 prism plane 4 basal
plane. Thus the type III AFP is essentially a non-basal ice plane
binder, which is in line with experimental evidence.58 We have
further characterized the nature of the AFP–ice complexes by
performing equilibrium simulations of the complexes corres-
Published on 14 August 2019. Downloaded by University of Perugia on 5/22/2023 8:46:37 AM.
correlates with the binding affinity data obtained from the PMF
calculations. We have also calculated the number of direct
hydrogen bonds between the AFP and the ice surface in three
different AFP–ice complexes from the simulation trajectories and
the results are shown in Fig. 4C. As is evident from the figure, the
basal plane has the highest ability to form direct hydrogen bonds
with the IBS and the pyramidal plane has the least affinity to form
direct hydrogen bonds with the IBS. The observed order of the
number of hydrogen bonds between the IBS and three different
ice planes is exactly opposite to the binding affinity order obtained

ponding to the PMF minima. Recent work on the binding of a
type I AFP and an insect AFP on the ice surface revealed that
AFPs essentially adsorb on the ice plane through a layer of
ordered water.48–50 The functional relevance of the water layer
has been critically analysed.49 Here, we have analysed the
nature of binding of the AFP on the ice surface, i.e., is it
adsorption of the AFP mediated through a layer of ordered
water or an AFP–ice hydrogen bonded complex? Distributions
of the number of water molecules sandwiched between the IBS
and the ice surface during the recognition of three different ice
planes are shown in Fig. 4B. The number of such water
molecules anchoring the IBS to the pyramidal plane is the
highest, followed by the prism plane, and it is least in the case
of basal plane binding (Fig. 4B). Interestingly, the order strongly
from the PMF calculations. Thus direct hydrogen bond formation
is not the factor that drives the AFP adsorption to a particular ice
plane, rather the number of interfacial water molecules that
anchor the AFP towards the ice surface dictates the binding
specificities. Hydrogen bonds may facilitate the complex forma-
tion but are not the deciding factor. Therefore, a type III AFP is
primarily adsorbed on the pyramidal plane through a layer of
ordered water. This mode of binding also has been reported for
type I AFPs and insect AFPs.48–50,85 However, the structure of a
type III AFP is distinctively different from those classes of AFPs.
Topologically and evolutionarily, a type III AFP does not bear any
structural resemblance to type I and insect AFPs.86
Further insight into the structure of the adsorbed complex
reveals that the IBS of the type III AFP is engulfed by a layer of

Fig. 4 (A) Potential of mean force (PMF) calculations for type III AFP adsorption on three different ice planes: pyramidal plane (black); prism plane (green);
and basal plane (pink). The results are represented as mean  S.D. (n = 4). Closer insights into the minima of the PMF profiles are shown in the inset. (B) The
distribution of interfacial water molecules between the IBS and ice surface during different ice plane adsorptions is shown. Water molecules are considered
to be interfacial when the oxygen atom of water is within 4.5 Å from the oxygen atoms of the ice surface and the IBS plane. (C) The number of hydrogen
bonds between the AFP and three different ice surfaces obtained from the equilibrium simulation of the adsorbed complexes are shown.

This journal is © the Owner Societies 2019 Phys. Chem. Chem. Phys., 2019, 21, 19298--19310 | 19303

PCCP
Published on 14 August 2019. Downloaded by University of Perugia on 5/22/2023 8:46:37 AM.
Fig. 5 3-D structure of the type III AFP–ice complex adsorbed on the
pyramidal plane. The AFP is shown in a pickle green cartoon representa-
tion and the IBS residues are represented in vdW mode. The ice layer is
shown in a stick representation. The interfacial water layer that anchors the
IBS on the ice surface is shown as a red transparent surface. Detailed
insight into the structure is shown in the inset (Side view: the water layer
that anchors the AFP on the ice surface is shown as red balls and the
hydrogen bonds between them are shown as a red dotted line; upper
view: the IBS of the type III AFP is shown as a pale green surface and the
hydration water around the IBS is shown as blue circles, the top surface of
the pyramidal plane adjacent to the IBS is also shown in stick mode).
water which enables the adherence of the AFP on the ice
surface (Fig. 5). Evident from the figure, most of the hydro-
phobic residues face the ice surface. Very few polar groups
orient towards the ice plane, which signifies fewer hydrogen
bonding possibilities between the IBS and the ice surface.
Further analysis of the water layer that anchors the AFP on
the ice surface reveals that there is a formation of water cages
around the IBS. Hydrophobic residues at the IBS induce water
ordering. This ordered water layer effectively adsorbs AFP on
the ice surface. In the final adsorbed state, the ice/water inter-
face and the ice surface complete the cage formation around
the IBS, which anchors the AFP to the pyramidal plane. The
water structure around the IBS closely resembles the surface
water arrangement of the pyramidal plane (Fig. 5, inset upper
view), as hydration water molecules (blue circles) are often
closely superimposable with the pyramidal surface water
arrangements (stick representation). This structural synergy
implies that upon adsorption a few intermediate water mole-
cules sandwiched between the IBS and the ice surface are
released to the bulk and as a result the ice surface becomes
part of the water cages framing the IBS. Close structural
matches between the hydration water around the IBS and the
pyramidal surface water arrangements rationalize the highest
binding affinity of the type III AFP for the pyramidal plane. The
basal surface is hexagonal and therefore there is a mismatch in
the water arrangement of the basal surface and the hydration
layer water structure of the type III AFP. Thus the spontaneity in
water cage formation has been reduced significantly, which
19304 | Phys. Chem. Chem. Phys., 2019, 21, 19298--19310
View Article Online
Paper
resulted in less effective adsorption complex formation on the
basal surface. Hudait et al. recently showed that the hydration
water around the IBS for a hyperactive insect AFP in solution
does not adopt a clathrate like structure, however, upon binding,
the hydration layer around the IBS adopts clathrate like
ordering.52 We have also observed that water cage formation
around the IBS is particularly pronounced in the adsorbed state
of the type III AFP on the pyramidal surface.
3.3. Effect of mutations on the conformational dynamics of
the type III AFP
Mutagenesis was often used to identify the ice binding hotspot
residues and also used to infer the mechanism of binding for type
III AFPs.21,55,87 Graether et al. summarised the activity of all the
previously reported type III AFP mutants and newly characterized
mutants in their work to provide a comprehensive understanding
of hotspot residues.88 From the TH data, it has been observed
that the T18N mutation severely affects the antifreeze activity
(90% loss in TH activity).87 Another substitution N14S causes a
75% loss in activity.87 These mutations were assumed to cause
steric interference during ice plane binding, which results in a
loss of hydrogen bonding between IBS and the ice surface.55 As a
control we have also considered the A16M mutation, which only
reduces the activity by 15%.35,89 Baardsnes et al. evaluated the role
of steric interactions in ice recognition by performing alanine
scanning mutation of several residues close to the proposed IBS.21
They found that single mutants L19A, V20A, and V41A showed
only a 20% loss of activity, however, double mutants showed more
than a 50% loss of activity. Among them, L19A/V41A showed a
75% loss of activity. Thus we have considered this particular
mutant in this study.21 We have critically evaluated the effects of
these mutations on the structure and binding ability of the type III
AFP. Notably, the chosen type III mutants showed a wide change
of TH activity and the nature of substitution is different from
each other. Choosing these widely different mutants with
varied activity profile provides a more robust validation of our
proposed adsorption model.
Initially, we have evaluated the effect of those mutations
on the conformational landscape of the protein and the IBS
dynamics using 500 ns of equilibrium simulation for each
system. The results are shown in Fig. 6.
RMSD is a global parameter that signifies structural deviations
of the protein during the simulation from the crystallographic
conformation and Rg is correlated with the size and shape of the
protein. 2-D scatter plots of the RMSD and Rg of wild-type and
mutant AFPs are shown in Fig. 6A. Here, each dot in the plot
represents a conformation of the protein sampled during the
simulation. Conformations of wild-type AFPs are confined in 2-D
space close to the crystallographic native form with RMSD
deviations of 1–2 Å. The T18N and N14S mutants show similar
distributions to the wild-type AFP. However, RMSF analysis
reveals that the fluctuations of the IBS are highest in the case
of the N14S mutant (Fig. 6B). Therefore, the observed loss of
activity of the N14S mutant can be partly attributed to the loss of
structural rigidity of the IBS. However, the IBS is highly stable for
the T18N mutant. Therefore, the loss of activity of the T18N mutant
This journal is © the Owner Societies 2019
 View Article Online

Paper PCCP
Published on 14 August 2019. Downloaded by University of Perugia on 5/22/2023 8:46:37 AM.

Fig. 6 Effects of several deleterious mutations on the dynamics of the type III AFP studied by using equilibrium simulations. (A) 2-D scatter plots of root
mean square deviations (RMSD) and radius of gyration (Rg) for wild type and various mutant type III AFPs obtained from equilibrium simulations are
shown. (B) Root mean square fluctuations (RMSF) of the ice binding plane residues in wild-type and mutant AFPs are also shown. (C) Cartoon
representations of the simulation average structure of type III and mutant AFPs coloured on the basis of their secondary structure (helix: red; b-sheet:
yellow; turn/coil: green) are shown. The IBS is shown within the red circle. Mutated residues are shown in a stick representation. The crystal structures of
the A16M (PDB id: 2MSI35) and T18N (PDB id: 9MSI88) mutants are shown in the inset.

is not due to the structural alteration of the protein due to
mutation. 2-D scatter plots for the A16M and L19A//V41A mutants
reveal that both the proteins sample conformations during the
simulation that are structurally distinct from the wild-type AFP
(Fig. 6A). However, the IBS is highly stable during the simulation
for the A16M mutant (Fig. 6B), which implies that the mutation
might alter the dynamics of the non-IBS regions rather than the
IBS. However, the double mutant shows slightly higher IBS fluctua-
tions during the simulation (Fig. 6B). The average simulation
structure reveals that a region of the IBS appears as b-bridge
(Fig. 6C, indicated by the circle) for the A16M and N14S mutants
rather than b-sheet, as exists in the wild-type AFP. Thus the
alteration in the structure of the protein induced by mutagenesis
is not the principal cause of the observed loss of activity for most of
the AFP mutants. However, high IBS fluctuations and a mild loss of
structural rigidity of the IBS for the N14S mutant might play a role
in the activity loss of the mutant to some extent.
3.4. Type III AFP mutants bind with the pyramidal plane with
reduced affinity
We have further investigated the role of these mutations
in the binding affinity with the ice plane using umbrella
sampling techniques. The potential of mean force for pyramidal
plane adsorption has been calculated for the wild-type and
different mutants and the results are shown in Fig. 7. Notably,
we have only considered the pyramidal plane binding affinity
since it has been shown before that the type III AFP binds to the
pyramidal plane with the highest affinity (Fig. 4A).
Mutations indeed reduce the binding affinity towards the
pyramidal plane and very interestingly the loss of activity is
strongly correlated with the loss of binding affinity (Fig. 7B),
except for the double mutant. Thus a mutation that reduces
the ice plane binding affinity essentially reduces the antifreeze
activity of the AFP, which indicates that appropriate ice plane
binding with considerable affinity is a critical event in the anti-
freezing activity of an AFP, at least for type III AFPs.
As we have shown that the wild-type AFP adsorbs on the
pyramidal plane with a layer of ordered water and hydrogen
bonds between the IBS and the ice surface further stabilize
the complex, so we have then separately analysed how each
of the mutations changes the ordering of the water layer
that anchors AFP to the ice surface and also the number
of IBS–ice surface hydrogen bonds. The results are shown in
Fig. 8.

This journal is © the Owner Societies 2019 Phys. Chem. Chem. Phys., 2019, 21, 19298--19310 | 19305
 View Article Online

PCCP Paper
Published on 14 August 2019. Downloaded by University of Perugia on 5/22/2023 8:46:37 AM.

Fig. 7 (A) Potential of mean force (PMF) profiles for pyramidal plane adsorption for the wild type and different type III AFP mutants are shown. Results are
represented as mean  S.D. (n = 4). (B) The correlation between the calculated binding free energy and experimentally observed thermal hysteresis
data21,35,55,87,89 (scaled with respect to the wild-type AFP activity) is shown.

Fig. 8 (A) The average number of water molecules anchoring the IBS with the pyramidal surface (black square) for the wild-type and different mutants
and the associated number fluctuations (pink square) are shown. Water molecules are considered as anchored when the oxygen atom of water is within
3.5 Å from the oxygen atoms of the ice surface and the IBS plane. (B) Distributions of IBS–ice surface hydrogen bonds for wild-type and mutant type III
AFP adsorption on the pyramidal plane are shown.

From the equilibrium simulations of the bound complexes
of the wild-type and mutant AFPs adsorbed on the pyramidal
plane, we have calculated the average number of interfacial
water molecules sandwiched between the IBS and the ice surface.
We have also characterized the average number fluctuations hwi of
those water molecules anchoring the wild-type/mutant AFPs on
the ice surface using the following expression:
 
1 X ðjNIW  hNIW ijÞ
hwi ¼ (1)
n hNIW i
hwi is the average number fluctuation, which is calculated by
averaging the instantaneous number fluctuations (wi) over
the entire simulation timescale. NIW is the number of water
molecules in between the IBS and the ice surface in a particular
time frame and hNIWi is the average number of such water
molecules calculated from the entire simulation trajectory. Low
fluctuations implicate the high stability of the water network
around the IBS.
The wild-type AFP adsorbs on the pyramidal plane with the
aid of a water layer constituted by 18–19 highly ordered water
molecules that form cages around the IBS. The water cage is
highly stable and associated with very little number fluctua-
tions (Fig. 8A). For the A16M mutant, the number of anchoring
water molecules decreases, while their associated fluctuations
increase marginally, which rationalizes the observed marginal
loss of binding affinity obtained from PMF calculations. Notably,
the A16M mutant is B85% active with respect to the wild-type.
In the case of the double mutant, there is a significant drop in

19306 | Phys. Chem. Chem. Phys., 2019, 21, 19298--19310 This journal is © the Owner Societies 2019
 View Article Online

Paper PCCP

the number of molecules of the anchoring water layer that

adsorbs the AFP on the ice surface and also its associated
fluctuation increases. In this mutant, bulkier side-chains are
mutated with small methyl groups that enable AFP to interact
more closely with the ice surface. It is clearly evident from the
PMF plot (Fig. 7A) that the double mutant interacts with the
pyramidal plane more closely in comparison to the wild-type.
Therefore, the number of interfacial water molecules sand-
wiched between the IBS and the ice surface decreases and
reduced complementarity increases the fluctuations of the water
Published on 14 August 2019. Downloaded by University of Perugia on 5/22/2023 8:46:37 AM.

cage. A significant disruption of the hydration water shell upon

ice adsorption can rationalize the high loss (B75%) in the
anti-freezing activity for the double mutant. However, since the
AFP mutant approaches closely to the ice surface, therefore
the number of IBS–ice surface hydrogen bonds increases
(Fig. 8B), which helps the mutant to gain in binding affinity
during adsorption. This explains the fact that L19A//V41A Fig. 9 O–O–O angle distributions of first hydration shell water molecules
exists as an outlier in the binding affinity and TH activity (within 5.5 Å) around the ice binding surface of the wild-type and different
correlation curve (Fig. 7B). Interestingly, for the N14S mutant AFP mutants are shown. The peak related to the tetrahedral water is
zoomed in on in the inset.
AFP, the number of anchoring water molecules is high and the
water network is also associated with much less fluctuation.
Therefore, the disruption of the hydration water shell around the hydration water around the IBS (Fig. 9). Interestingly, the
the IBS is not the reason for the loss of binding affinity. order of reduction in tetrahedrality of the hydration water in
However, there is a significant loss of hydrogen bonding the wild-type and AFP mutants (wild-type 4 A16M 4 L19A//
between the IBS and the ice surface (Fig. 8B). Notably, we V41A 4 N14S 4 T18N) is strongly correlated with the order of
have demonstrated that for the N14S mutant, the IBS lost the binding affinity obtained from the PMF calculations.
its rigidity. Thus structural alterations play a significant
role in the loss of binding affinity as well as its activity. Very 3.5. Proposed mechanism
fascinating results have been observed for the T18N mutant. Based on the detailed investigations on the mechanism of ice
The mutant AFP forms more direct hydrogen bonds with the plane adsorption of the type III AFP and its different mutants,
ice surface than the wild-type (Fig. 8B), yet the mutant has structured hydration water layer mediated ice recognition by
the least binding affinity. The loss of binding affinity is the type III AFP is proposed. The mechanism is schematically
completely associated with the loss of water cage formation shown in Fig. 10.
during pyramidal plane adsorption. The number of anchoring
water molecules decreases drastically and also its associated
fluctuations are very high, which signifies the least stability of
the water cages that anchor the T18N mutant AFP to the ice
surface (Fig. 8A). Our findings rationalize the fact that the hydration
water around the thr18 residue of the wild type AFP is highly
tetrahedral, evident from sum frequency generation spectroscopy.59
The result also clearly signifies hydrogen bond formation with the
ice surface is not the driving force for binding, rather ordering of
the hydration water around the IBS and proper complementarity
between the hydration water structure and ice/water interface
structure drives the AFP–ice adsorption complex formation. Hydro-
gen bonds that complement and stabilize the water network
around the IBS favour the complex formation further.
In fact, the loss of stability of the water cages around the
IBS during adsorption to the pyramidal surface is strongly
correlated with the inherent loss of hydration shell ordering
in different AFP mutants. Comparisons of the O–O–O angle
distributions around the IBS among wild-type and mutant
Fig. 10 Schematic outline of the adsorption mechanism of the type III
AFPs show that the frequency of the higher angle peak corres-
AFP on the ice surface. The AFP and the ice layer are rendered in cartoon
ponding to the tetrahedrally ordered water is lowest in the case and stick representations, respectively. The IBS is shown in a surface
of the T18N mutant. Also, the peak is shifted to a lower angle representation and the ordered hydration water cages around the IBS
value, which indicates the lowering of tetrahedral ordering in are shown as red sticks.

This journal is © the Owner Societies 2019 Phys. Chem. Chem. Phys., 2019, 21, 19298--19310 | 19307

PCCP
The ice binding surface (IBS) of the type III AFP is mostly flat
and hydrophobic. Hydrophobic hydration at low temperature
induces the ordering of water molecules in the hydration shell
around the IBS. There is even the formation of water cage
structures BTm. The close synergy between the water arrangements
in the water cages around the IBS and the surface water
arrangements of the pyramidal plane drives the AFP to adhere
on the pyramidal ice surface. In the adsorbed complex, the ice
surface becomes part of the water cages, which effectively
adsorbs the AFP on the ice surface. A specific mutation of the
Published on 14 August 2019. Downloaded by University of Perugia on 5/22/2023 8:46:37 AM.
IBS residues that disrupts the ordering of the hydration layer
around the IBS leads to less effective adsorption. Disruption of
the water ordering around the IBS results in a large reduction
in the binding affinity of the mutant AFPs which can’t be
compensated for by direct hydrogen bond formation with the
ice surface. The principal driving force for this hydration water
mediated ice recognition is hydrophobic hydration, which
becomes highly significant in the low temperature regime.
It is noteworthy that the working temperature of AFPs is around
freezing temperature. A strong linear correlation between the
binding affinity and activity data in several AFP mutants
suggests that the binding to a specific ice plane dictates the
observed thermal hysteresis (TH) activity.
4. Conclusions
The present study explores the mechanism of ice/water inter-
face recognition by a type III AFP. Previous studies on type I and
hyperactive insect AFPs suggested an anchored clathrate water
mediated ice adsorption mechanism.48–52 The IBSs of both
hyperactive insect AFPs and type I AFPs contain equally spaced
periodic threonine residues, known as Thr ladders. Methyl
side-chain and hydroxyl groups play a crucial role in stabilizing
the anchored clathrate water array which anchors the AFP to
the ice surface by forming co-operative dual hydrogen bonds.49
However, type III AFPs are globular proteins and structurally
and topologically a type III AFP is completely different from
type I and insect AFPs. Importantly, the IBS does not contain
any repeat of threonine or any other related amino acids.
Rather, the IBS is comparatively flat and more hydrophobic
compared to the insect AFPs. Here, we have observed an IBS
induced ordered hydration water mediated ice/water interface
recognition mechanism. The flat hydrophobic IBS induces
water structure formation in the freezing temperature regime
which drives the AFP to adhere to the ice/water interface of the
pyramidal plane and to some extent to the prism plane too.
Structural synergies between the water cages formed around
the IBS and the interfacial water of the pyramidal plane highly
assist the adsorption. This observation justifies the fact that
type III AFPs are essentially non-basal binders supported by the
ice-etching study.58
The ice/water interface recognition mechanism by type III
AFPs is surprising in the context of the currently accepted three
step model for protein adsorption on solid surfaces where
anchoring by the hydrophilic residues to the interfacial water
19308 | Phys. Chem. Chem. Phys., 2019, 21, 19298--19310
View Article Online
Paper
layer of the solid surface plays a central role. A type III AFP
recognizes ice using its flat hydrophobic surface. Our study
shows that hydrophobic hydration is the primary driving force
for the AFP–ice complex formation, not the direct hydrogen
bond formation between the IBS and the ice surface. Mutations
that interrupt the hydration shell water ordering essentially
lead to less efficient adsorption on the ice surface, which in
turn greatly reduces the activity.
Comparisons of our results with the ice binding mechanism
of other classes of AFPs, type I, type II90 and insect AFPs,48,49,53
indicate a universal mechanism of ice recognition among
different classes of AFPs: ordered hydration layer mediated
ice recognition. However, the origin of such water ordering
around the IBS is completely different in different types of AFPs
and strongly depends on the sequence and topologies of the IBS.
The evolution of different classes of AFPs was independent,32 which
essentially leads to such structural diversities in the IBS among
different AFPs.86 As a result the ice binding affinities and activities
of different classes of AFPs greatly vary from weak, moderate to
hyperactive. However, the core recognition mechanism remains
preserved through the process of natural selection.
Conflicts of interest
There are no conflicts to declare.
Acknowledgements
The authors sincerely acknowledge financial assistance from
the Department of Science and Technology, Government of India
(SERB grant EMR/2016/001333). The authors also gratefully
acknowledge the central supercomputing facility (CRAY) at the
Indian Association for the Cultivation of Science, Kolkata.
References
1 M. M. Orosco, C. Pacholski and M. J. Sailor, Nat. Nanotech-
nol., 2009, 4, 255.
2 T. Fischer, A. Agarwal and H. Hess, Nat. Nanotechnol., 2009,
4, 162.
3 H. Im, X.-J. Huang, B. Gu and Y.-K. Choi, Nat. Nanotechnol.,
2007, 2, 430.
4 A. E. Nel, L. Mädler, D. Velegol, T. Xia, E. M. V. Hoek,
P. Somasundaran, F. Klaessig, V. Castranova and M. Thompson,
Nat. Mater., 2009, 8, 543.
5 C. Wu, M. Chen, A. A. Skelton, P. T. Cummings and
T. Zheng, ACS Appl. Mater. Interfaces, 2013, 5, 2567–2579.
6 Y. Cheng, L.-D. Koh, D. Li, B. Ji, Y. Zhang, J. Yeo, G. Guan,
M.-Y. Han and Y.-W. Zhang, ACS Appl. Mater. Interfaces,
2015, 7, 21787–21796.
7 K. Yao, P. Tan, Y. Luo, L. Feng, L. Xu, Z. Liu, Y. Li and
R. Peng, ACS Appl. Mater. Interfaces, 2015, 7, 12270–12277.
8 Q. Q. Hoang, F. Sicheri, A. J. Howard and D. S. C. Yang,
Nature, 2003, 425, 977.
This journal is © the Owner Societies 2019

Paper
9 X. Sun, Z. Feng, T. Hou and Y. Li, ACS Appl. Mater. Interfaces,
2014, 6, 7153–7163.
10 Y. Yu, H. Sun, K. Gilmore, T. Hou, S. Wang and Y. Li,
ACS Appl. Mater. Interfaces, 2017, 9, 32452–32462.
11 A. M. Sultan, Z. C. Westcott, Z. E. Hughes, J. P. Palafox-
Hernandez, T. Giesa, V. Puddu, M. J. Buehler, C. C. Perry and
T. R. Walsh, ACS Appl. Mater. Interfaces, 2016, 8, 18620–18630.
12 A. A. Skelton, T. Liang and T. R. Walsh, ACS Appl. Mater.
Interfaces, 2009, 1, 1482–1491.
13 L. Vroman, Nature, 1962, 196, 476.
Published on 14 August 2019. Downloaded by University of Perugia on 5/22/2023 8:46:37 AM.
14 M. J. Penna, M. Mijajlovic and M. J. Biggs, J. Am. Chem. Soc.,
2014, 136, 5323–5331.
15 J. Yu, M. L. Becker and G. A. Carri, Langmuir, 2012, 28,
1408–1417.
16 P. Harder, M. Grunze, R. Dahint, G. M. Whitesides and
P. E. Laibinis, J. Phys. Chem. B, 1998, 102, 426–436.
17 S. I. Jeon, J. H. Lee, J. D. Andrade and P. G. De Gennes,
J. Colloid Interface Sci., 1991, 142, 149–158.
18 I. Szleifer, Biophys. J., 1997, 72, 595–612.
19 L. Li, S. Chen, J. Zheng, B. D. Ratner and S. Jiang, J. Phys.
Chem. B, 2005, 109, 2934–2941.
20 W. Norde, Colloids Surf., B, 2008, 61, 1–9.
21 J. Baardsnes and P. L. Davies, Biochim. Biophys. Acta, Proteins
Proteomics, 2002, 1601, 49–54.
22 W. Zhang and R. A. Laursen, J. Biol. Chem., 1998, 273,
34806–34812.
23 Q. Ye, E. Leinala and Z. Jia, Acta Crystallogr., Sect. D: Biol.
Crystallogr., 1998, 54, 700–702.
24 J. H. Kim, H. J. Lee, B. Y. Hur, W. C. Lee, S.-H. Park and
B.-W. Koo, Mar. Drugs, 2017, 15, 27.
25 W. Gronwald, M. C. Loewen, B. Lix, A. J. Daugulis,
F. D. Sönnichsen, P. L. Davies and B. D. Sykes, Biochemistry,
1998, 37, 4712–4721.
26 N. Xiao, K. Suzuki, Y. Nishimiya, H. Kondo, A. Miura,
S. Tsuda and T. Hoshino, FEBS J., 2009, 277, 394–403.
27 S. P. Graether, M. J. Kuiper, S. M. Gagne, V. K. Walker, Z. Jia,
B. D. Sykes and P. L. Davies, Nature, 2000, 406, 325–328.
28 M. M. Tomczak, C. B. Marshall, J. A. Gilbert and P. L. Davies,
Biochem. Biophys. Res. Commun., 2003, 311, 1041–1046.
29 S. Y. Gauthier, A. J. Scotter, F.-H. Lin, J. Baardsnes, G. L.
Fletcher and P. L. Davies, Cryobiology, 2008, 57, 292–296.
30 Z. Jia and P. L. Davies, Trends Biochem. Sci., 2002, 27, 101–106.
31 A. L. Devries, Science, 1971, 172, 1152–1155.
32 C.-H. C. Cheng, Curr. Opin. Genet. Dev., 1998, 8, 715–720.
33 A. L. Devries and Y. Lin, Biochim. Biophys. Acta, Protein
Struct., 1977, 495, 388–392.
34 D. Wen and R. A. Laursen, Biophys. J., 1992, 63, 1659–1662.
35 C. I. DeLuca, P. L. Davies, Q. Ye and Z. Jia, J. Mol. Biol., 1998,
275, 515–525.
36 D. S. C. Yang, W.-C. Hon, S. Bubanko, Y. Xue, J. Seetharaman,
C. L. Hew and F. Sicheri, Biophys. J., 1998, 74, 2142–2151.
37 Y. Celik, R. Drori, N. Pertaya-Braun, A. Altan, T. Barton,
M. Bar-Dolev, A. Groisman, P. L. Davies and I. Braslavsky,
Proc. Natl. Acad. Sci. U. S. A., 2013, 110, 1309–1314.
38 D. R. Nutt and J. C. Smith, J. Am. Chem. Soc., 2008, 130,
13066–13073.
This journal is © the Owner Societies 2019
View Article Online
PCCP
39 K. Meister, S. Ebbinghaus, Y. Xu, J. G. Duman, A. DeVries,
M. Gruebele, D. M. Leitner and M. Havenith, Proc. Natl.
Acad. Sci. U. S. A., 2013, 110, 1617–1622.
40 S. Ebbinghaus, K. Meister, B. Born, A. L. DeVries, M. Gruebele
and M. Havenith, J. Am. Chem. Soc., 2010, 132, 12210–12211.
41 S. Ebbinghaus, K. Meister, M. B. Prigozhin, A. L. DeVries,
M. Havenith, J. Dzubiella and M. Gruebele, Biophys. J., 2012,
103, L20–L22.
42 S. S. Mallajosyula, K. Vanommeslaeghe and A. D. MacKerell,
J. Phys. Chem. B, 2014, 118, 11696–11706.
43 G. Todde, C. Whitman, S. Hovmöller and A. Laaksonen,
J. Phys. Chem. B, 2014, 118, 13527–13534.
44 G. Todde, S. Hovmöller and A. Laaksonen, J. Phys. Chem. B,
2015, 119, 3407–3413.
45 N. Pertaya, C. B. Marshall, C. L. DiPrinzio, L. Wilen, E. S.
Thomson, J. S. Wettlaufer, P. L. Davies and I. Braslavsky,
Biophys. J., 2007, 92, 3663–3673.
46 N. Pertaya, C. B. Marshall, Y. Celik, P. L. Davies and
I. Braslavsky, Biophys. J., 2008, 95, 333–341.
47 R. Drori, Y. Celik, P. L. Davies and I. Braslavsky, J. R. Soc.,
Interface, 2014, 11, 20140526.
48 S. Chakraborty and B. Jana, Phys. Chem. Chem. Phys., 2017,
19, 11678–11689.
49 S. Chakraborty and B. Jana, Langmuir, 2017, 33, 7202–7214.
50 S. Chakraborty and B. Jana, J. Phys. Chem. B, 2018, 122,
3056–3067.
51 A. Hudait, N. Odendahl, Y. Qiu, F. Paesani and V. Molinero,
J. Am. Chem. Soc., 2018, 140, 4905–4912.
52 A. Hudait, D. R. Moberg, Y. Qiu, N. Odendahl, F. Paesani and
V. Molinero, Proc. Natl. Acad. Sci. U. S. A., 2018, 115, 8266.
53 K. Meister, C. J. Moll, S. Chakraborty, B. Jana, A. L. DeVries,
H. Ramløv and H. J. Bakker, J. Chem. Phys., 2019, 150, 131101.
54 C. P. Garnham, R. L. Campbell and P. L. Davies, Proc. Natl.
Acad. Sci. U. S. A., 2011, 108, 7363–7367.
55 Z. Jia, C. I. DeLuca, H. Chao and P. L. Davies, Nature, 1996,
384, 285–288.
56 D. Verreault, S. Alamdari, S. J. Roeters, R. Pandey,
J. Pfaendtner and T. Weidner, Phys. Chem. Chem. Phys.,
2018, 20, 26926–26933.
57 C. C. Cheng and A. L. DeVries, The role of antifreeze
glycopeptides and peptides in the freezing avoidance of cold-
water fish, Berlin, Heidelberg, 1991, pp. 1–14.
58 A. A. Antson, D. J. Smith, D. I. Roper, S. Lewis, L. S. D. Caves,
C. S. Verma, S. L. Buckley, P. J. Lillford and R. E. Hubbard,
J. Mol. Biol., 2001, 305, 875–889.
59 Z. F. Brotzakis, I. K. Voets, H. J. Bakker and P. G. Bolhuis,
Phys. Chem. Chem. Phys., 2018, 20, 6996–7006.
60 N. Smolin and V. Daggett, J. Phys. Chem. B, 2008, 112, 6193–6202.
61 C. Yang and K. A. Sharp, Biophys. Chem., 2004, 109, 137–148.
62 K. Meister, S. Strazdaite, A. L. DeVries, S. Lotze, L. L. C.
Olijve, I. K. Voets and H. J. Bakker, Proc. Natl. Acad. Sci.
U. S. A., 2014, 111, 17732–17736.
63 J. Grabowska, A. Kuffel and J. Zielkiewicz, J. Chem. Phys.,
2016, 145, 075101.
64 Y. Xu, A. Bäumer, K. Meister, C. G. Bischak, A. L. DeVries,
D. M. Leitner and M. Havenith, Chem. Phys. Lett., 2016, 647, 1–6.
Phys. Chem. Chem. Phys., 2019, 21, 19298--19310 | 19309

PCCP
65 S. Mahatabuddin, D. Fukami, T. Arai, Y. Nishimiya,
R. Shimizu, C. Shibazaki, H. Kondo, M. Adachi and
S. Tsuda, Proc. Natl. Acad. Sci. U. S. A., 2018, 115, 5456.
66 T. Sun, S. Y. Gauthier, R. L. Campbell and P. L. Davies,
J. Phys. Chem. B, 2015, 119, 12808–12815.
67 M. M. Teeter, Proc. Natl. Acad. Sci. U. S. A., 1984, 81, 6014.
68 R. Talon, N. Coquelle, D. Madern and E. Girard, Front.
Microbiol., 2014, 5, 66.
69 D. Van Der Spoel, E. Lindahl, B. Hess, G. Groenhof, A. E.
Mark and H. J. C. Berendsen, J. Comput. Chem., 2005, 26,
Published on 14 August 2019. Downloaded by University of Perugia on 5/22/2023 8:46:37 AM.
1701–1718.
70 E. Lindahl, B. Hess and D. van der Spoel, J. Mol. Model.,
2001, 7, 306–317.
71 H. J. C. Berendsen, D. van der Spoel and R. van Drunen,
Comput. Phys. Commun., 1995, 91, 43–56.
72 G. A. Kaminski, R. A. Friesner, J. Tirado-Rives and W. L.
Jorgensen, J. Phys. Chem. B, 2001, 105, 6474–6487.
73 W. L. Jorgensen and J. Tirado-Rives, J. Am. Chem. Soc., 1988,
110, 1657–1666.
74 W. L. Jorgensen, D. S. Maxwell and J. Tirado-Rives, J. Am.
Chem. Soc., 1996, 118, 11225–11236.
75 H. J. C. Berendsen, J. R. Grigera and T. P. Straatsma, J. Phys.
Chem., 1987, 91, 6269–6271.
76 T. Bryk and A. D. J. Haymet, J. Chem. Phys., 2002, 117,
10258–10268.
19310 | Phys. Chem. Chem. Phys., 2019, 21, 19298--19310
View Article Online
Paper
77 W. Humphrey, A. Dalke and K. Schulten, J. Mol. Graphics,
1996, 14, 33–38.
78 G. Bussi, D. Donadio and M. Parrinello, J. Chem. Phys., 2007,
126, 014101.
79 T. Darden, D. York and L. Pedersen, J. Chem. Phys., 1993, 98,
10089–10092.
80 D. Schneidman-Duhovny, Y. Inbar, R. Nussinov and
H. J. Wolfson, Nucleic Acids Res., 2005, 33, W363–W367.
81 J. S. Hub, B. L. de Groot and D. van der Spoel, J. Chem.
Theory Comput., 2010, 6, 3713–3720.
82 C. P. Garnham, A. Natarajan, A. J. Middleton, M. J. Kuiper,
I. Braslavsky and P. L. Davies, Biochemistry, 2010, 49, 9063–9071.
83 S. Parui and B. Jana, J. Phys. Chem. B, 2017, 121, 7016–7026.
84 S. Parui and B. Jana, J. Phys. Chem. B, 2018, 122, 9827–9839.
85 J. Grabowska, A. Kuffel and J. Zielkiewicz, Phys. Chem. Chem.
Phys., 2018, 20, 25365–25376.
86 S. Chakraborty and B. Jana, Proc. Indian Natl. Sci. Acad.,
2019, 85, 169–187.
87 H. Chao, C. I. DeLuca, P. L. Davies, B. D. Sykes and
F. D. Sönnichsen, Protein Sci., 1994, 3, 1760–1769.
88 S. P. Graether, C. I. DeLuca, J. Baardsnes, G. A. Hill, P. L.
Davies and Z. Jia, J. Biol. Chem., 1999, 274, 11842–11847.
89 C. I. DeLuca, H. Chao, F. D. Sönnichsen, B. D. Sykes and
P. L. Davies, Biophys. J., 1996, 71, 2346–2355.
90 S. Chakraborty and B. Jana, Metallomics, 2019, 11, 1387–1400.
This journal is © the Owner Societies 2019