Article pubs.acs.org/JPCB
Influence of Antifreeze Proteins on the Ice/Water Interface Guido Todde,† Sven Hovmöller,† and Aatto Laaksonen*,†,‡,§ †
Department of Material and Environmental Chemistry, Arrhenius Laboratory, Stockholm University, 10691 Stockholm, Sweden Science for Life Laboratory, 17121 Solna, Sweden § Stellenbosch Institute of Advanced Study (STIAS), Wallenberg Research Centre at Stellenbosch University, 7600 Stellenbosch, South Africa ‡
S Supporting Information *
ABSTRACT: Antifreeze proteins (AFP) are responsible for the survival of several species, ranging from bacteria to fish, that encounter subzero temperatures in their living environment. AFPs have been divided into two main families, moderately and hyperactive, depending on their thermal hysteresis activity. We have studied one protein from both families, the AFP from the snow flea (sfAFP) and from the winter flounder (wfAFP), which belong to the hyperactive and moderately active family, respectively. On the basis of molecular dynamics simulations, we have estimated the thickness of the water/ice interface for systems both with and without the AFPs attached onto the ice surface. The calculation of the diffusion profiles along the simulation box allowed us to measure the interface width for different ice planes. The obtained widths clearly show a different influence of the two AFPs on the ice/water interface. The different impact of the AFPs here studied on the interface thickness can be related to two AFPs properties: the protein hydrophobic surface and the number of hydrogen bonds that the two AFPs faces form with water molecules.
■
INTRODUCTION Antifreeze proteins (AFP) belong to a family of proteins responsible for the freezing point depression of physiological fluids. They have been found in a great variety of life forms such as insects,1−5 fish,6−9 plants,10 and bacteria.11,12 The AFPs allow all of these organisms to survive in subzero environments. The depression of the freezing point of the solution where AFPs are dissolved is a noncolligative property. AFPs lower the freezing point of a solution without affecting its melting point. This property is called thermal hysteresis (TH), which represents the difference between the melting and the freezing point of the solution. The mechanism behind the TH activity of AFPs is assumed to be due to the Gibbs−Thomson (Kelvin) effect. AFPs “bind” to one or more ice crystal surfaces, creating microcurvature13 of the ice surface, which hinders further growth of ice. The term “bind” has been questioned during the years. First, the interaction was proposed to be a of receptor−ligand kind between the AFPs and the ice surface.14 Recently, new studies have pointed the attention to the ice/water interface. Wierzbicki et al.15 proposed that the protein is adsorbed onto the ice/water interface, arguing that a receptor−ligand type of interaction would not take into account the interface between the two water phases. The adsorption of AFPs onto the ice/ water interface was first proposed by Haymet et al.16 and then reported also in other studies.17,18 Many theoretical studies19−25 have been carried out on the ice/water interface and its properties, but to the best of our knowledge none of them include an AFP molecule. From the ice/water interface studies, it was reported that the four ice planes considered in those © XXXX American Chemical Society
studies show slightly different thickness of the interface. The different interfaces were estimated to be 10−20 Å thick22−25 going from the basal interface (the thickest) to the bipyramidal interface (the thinnest).23 AFPs show very different amino acid sequences and 3-D structures, suggesting a convergent evolution that gave different answers (in terms of protein structure) to the same problem of adaptability to an extreme environment. All AFPs share one common feature. They are all amphipathic having one hydrophilic and one hydrophobic surface, which probably is the origin of their thermal hysteresis activity. In terms of their activity, AFPs have been divided into two main categories: moderately active and hyperactive. In the work of Scotter et al.,26 different AFPs were grouped as follows: to the moderately active group belong, for example, AFPs from the winter flounder (wfAFP, isoform HPLC-6), the pure calcium-dependent type II of atlantic herring (ahAFP), and the shorthorn sculpin (ssAFP); while to the hyperactive group belong, for example, AFPs from the spruce budworm (sbwAFP), the Tenebrio molitor (TmAFP), the Antarctic bacterium Marinomonas primoryensis (MpAFP), and the snow flea (sfAFP). In this work, we have studied, through MD simulations, the influence of two AFPs on the ice/water interface of four different ice planes. We have studied one protein belonging to the hyperactive group (the sfAFP) and the most studied AFP belonging to the moderately active group (the wfAFP). Received: December 1, 2014 Revised: January 21, 2015
A
DOI: 10.1021/jp5119713 J. Phys. Chem. B XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry B
wfAFP), and PDB ID 1UBQ49 (ubiquitin). The potential energies of the both proteins were minimized in vacuum for 1000 steps with a Steepest Descent algorithm. The resulting structures were then used for the simulations at the ice/water interfaces. AFP/Ice/Liquid Water. The structures, obtained after the equilibration protocol followed for the ice/liquid water systems, were used to insert the protein in the two-phase water systems. The protein coordinates, obtained from the protein energy minimization in vacuum, were translated and rotated until the flat hydrophobic face of the protein was parallel to the ice plane. The starting configurations are presented in Figure 1. The starting box dimensions are equal to those of the ice/ liquid water systems. The empty volume needed for inserting the protein was obtained by removing all water molecules within 3 Å from the protein (∼450 molecules for sfAFP and ∼250 molecules for the wfAFP). The water potential energy was then minimized for 1000 steps with a Steepest Descent algorithm keeping the protein coordinates fixed. Thereafter, a 50 ps molecular dynamics simulation was performed by imposing constraints on all protein bonds. At this point, 10 ns MD simulations, without any constraints on the proteins atoms, were carried out for all systems at four different temperatures, 224, 225, 226, and 227 K. The trajectories were saved every 1 ps. Diffusion Coefficient and Other Properties. The diffusion coefficient profile was calculated for all four ice/ water interfaces with and without the AFPs bound to the ice surface. It was calculated for all water molecules within intervals 1 Å thick. The interval was then shifted by 0.5 Å along the longest box direction (depending on the simulated interface, see Table 1). The root-mean-square displacement of all molecules within all intervals was computed along the 10 ns trajectory, which was divided into 14 segments of 700 ps each. The diffusion coefficient D was calculated from the slope of the root-mean-square displacement of all water molecules within each interval and averaged over the entire simulation time. The value of the diffusion coefficient was estimated using the Einstein relation:50−52
Moreover, we have simulated a non-AFP protein as a control. We have estimated the ice/water interface thickness for systems with and without the AFPs adsorbed onto it at different temperatures. To estimate the interface thickness, we have computed the diffusion profile perpendicular to the ice plane direction for all four ice planes.
■
METHODS AND MODELS Force Field. All of the molecular dynamics simulations were carried out using Gromacs 4.5.427−30 simulation software. For the proteins the OPLS-AA force field31−37 was used, while water molecules were modeled with the TIP4P model.38 NPT ensemble was adopted for all simulations. Pressure was set to 1 atm and controlled by the Parrinello−Rahman barostat.39,40 Temperature was controlled by the Nose−Hoover thermostat.41,42 Cut-off radius for both Coulombic interactions (calculated using the PME method43,44) and Lennard-Jones interactions was 10 Å. All bonds involving hydrogen atoms were treated by the LINCS algorithm.45 Ice/Liquid Interface. The starting ice slabs were downloaded from the Sonnichsen lab.46 The starting ice slab was translated and replicated to create a simulation box with enough bulk water in both phases (liquid and solid). The final dimension of the boxes depends on the simulated ice plane. The box dimensions and the number of water molecules in both phases are shown in Table 1. Table 1. Box Dimensions and Number of Water Molecules ice plane prism secondary prism bipyramidal basal
box dimensions (Å)
water molecules (liquid phase)
water molecules (ice)
66 × 67 × 185 61 × 217 × 59
12 960 12 288
12 960 12 288
181 × 67 × 75 68 × 67 × 165
14 724 11 880
14 076 11 880
The simulation protocol is from our previous work.47 The potential energy of the initial configurations were minimized for 1000 steps with a Steepest Descent algorithm. Three short MD simulations (50 ps) were thereafter sequentially performed at increasing temperatures (100, 150, and 200 K) in an NPT ensemble imposing the position restraints on all water oxygen atoms. The system was thereafter simulated for 100 ps at 300 K allowing the water molecules in one-half of the box to move freely. In this way, we obtained all four ice planes at the interface with water. The final configuration was used as the starting structure for the ice/liquid water simulations. For all four ice planes, a 10 ns MD was performed in an NPT ensemble storing the configurations every 1 ps. Five different temperatures were chosen for the trajectory production: 224, 225, 226, 227, and 228 K. The temperatures range from the melting point of hexagonal ice (Ih) obtained in our previous study47 to temperature 4 K below that point. Proteins Energy Minimization. The starting coordinates of the sfAFP were obtained from the crystal structure solved by X-ray diffraction at 0.98 Å resolution. The starting coordinates of the type I AFP were obtained from the crystal structure solved by X-ray diffraction at 1.50 Å resolution. The starting coordinates of the ubiquitin were obtained from the crystal structure solved by X-ray diffraction at 1.80 Å resolution. The three sets of coordinates were obtained from the Protein Data Bank,48 PDB ID 2PNE5 (sfAFP), PDB ID 1WFB6 (type I AFP,
D = lim
t →∞
⟨[r(t ) − r(0)]2 ⟩ 6kT
The solvent accessible surface (SAS) is computed using GROMACS tool g_sas. Atoms are classified as hydrophilic and hydrophobic according to their partial charge. Hydrophobic atoms have a partial charge between −0.2 and +0.2. Hydrogen bonds (HB) are defined by a donor−acceptor cutoff distance of 3.4 Å and a ∠DHA angle of 45°.
■
RESULTS AND DISCUSSION Ice/Liquid Interface. The evolution of each system was followed by the total energy profile. For the TIP4P water model, a melting temperature (Tm) of 228−229 K was estimated by previous studies.47,53 Following the drift of the energy profile (see Figure 1), we found a Tm of 228 K for the secondary prism and bipyramidal ice plane and between 227 and 228 K for the prism and the basal plane. The Tm is evaluated by the drift change of the energy profile.53 The Tm is obtained at the temperature when the drift passes from positive (ice is melting) to negative (ice is growing). The obtained results are satisfactory and within an error of ±2 K. B
DOI: 10.1021/jp5119713 J. Phys. Chem. B XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry B
Figure 1. Starting configuration of the protein/ice/water systems: (a) sfAFP; (b) wfAFP; (c) ubiquitin. In all systems, water (liquid) is in red, ice is in cyan, and the protein is in green.
order of the interfaces and their widths are found in good agreement with previous estimations of ice/water interface width of ∼10−20 Å.22−25 Also, the interface span (2 Å) is consistent with the estimate of Hayward and Haymet.23,24 AFPs at the Ice/Liquid Interface. The same procedure for the estimation of the ice/water interface thickness was applied also to the systems containing the proteins. The interface widths for all ice planes are presented in Table 3. It can be easily seen that the interface thickness has become wider for all ice planes for the two AFPs. To evaluate the influence of the two AFPs, we have reported the width difference as a percentage, calculated with respect to the values of the ice/ water systems (see Table 2). The two AFPs affect the interface width to different extents. The sfAFP leads to an increase of the interface width of ∼25−40%, while for the wfAFP the difference is estimated to be about 10−20%. The sfAFP shows a width increase very similar for all ice planes except for the basal plane, which is slightly less broadened. The wfAFP instead induces an increase of the interface for the secondary
All simulation boxes were built to contain enough bulk water molecules. The box axis containing the ice/water interface was set to be the longest box dimension. As shown in Figure 2, the periodic boundary conditions create two ice/water interfaces for each box. In this way, we could get twice as much diffusion data for each ice plane. The thickness of the interface was estimated from the diffusion profiles (see Supporting Information Figure S1) through the 10−90% width.19−21,23,24 The 10−90% width can be used to estimate the thickness of any interface. It is defined as the length over which a certain property varies from 10% to 90% of the bulk value (liquid water in our case) when passing across the interface from the solid to the liquid phase. The widths of the different ice planes are presented in Table 2. The values are obtained as averages over the five different temperatures set in the simulations for the two interfaces of each box. The bulk water values were calculated from the bulk water region of each box (Figure 3). The estimated widths going from the thickest to the thinnest interface are basal, prism, bipyramidal, secondary prism. The C
DOI: 10.1021/jp5119713 J. Phys. Chem. B XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry B
Figure 2. Evolution of the total energy for the four ice planes at different temperatures.
are consistent with the classification of the two AFPs given by Scotter et al.26 They assign the sfAFP to the hyperactive family, while the wfAFP belong to the moderately active group. SfAFP has a thermal hysteresis activity of 1.0 °C at a concentration of 0.02 mM, while the wfAFP produces
Influence of Antifreeze Proteins on the Ice/Water Interface Guido Todde,† Sven Hovmöller,† and Aatto Laaksonen*,†,‡,§ †
Department of Material and Environmental Chemistry, Arrhenius Laboratory, Stockholm University, 10691 Stockholm, Sweden Science for Life Laboratory, 17121 Solna, Sweden § Stellenbosch Institute of Advanced Study (STIAS), Wallenberg Research Centre at Stellenbosch University, 7600 Stellenbosch, South Africa ‡
S Supporting Information *
ABSTRACT: Antifreeze proteins (AFP) are responsible for the survival of several species, ranging from bacteria to fish, that encounter subzero temperatures in their living environment. AFPs have been divided into two main families, moderately and hyperactive, depending on their thermal hysteresis activity. We have studied one protein from both families, the AFP from the snow flea (sfAFP) and from the winter flounder (wfAFP), which belong to the hyperactive and moderately active family, respectively. On the basis of molecular dynamics simulations, we have estimated the thickness of the water/ice interface for systems both with and without the AFPs attached onto the ice surface. The calculation of the diffusion profiles along the simulation box allowed us to measure the interface width for different ice planes. The obtained widths clearly show a different influence of the two AFPs on the ice/water interface. The different impact of the AFPs here studied on the interface thickness can be related to two AFPs properties: the protein hydrophobic surface and the number of hydrogen bonds that the two AFPs faces form with water molecules.
■
INTRODUCTION Antifreeze proteins (AFP) belong to a family of proteins responsible for the freezing point depression of physiological fluids. They have been found in a great variety of life forms such as insects,1−5 fish,6−9 plants,10 and bacteria.11,12 The AFPs allow all of these organisms to survive in subzero environments. The depression of the freezing point of the solution where AFPs are dissolved is a noncolligative property. AFPs lower the freezing point of a solution without affecting its melting point. This property is called thermal hysteresis (TH), which represents the difference between the melting and the freezing point of the solution. The mechanism behind the TH activity of AFPs is assumed to be due to the Gibbs−Thomson (Kelvin) effect. AFPs “bind” to one or more ice crystal surfaces, creating microcurvature13 of the ice surface, which hinders further growth of ice. The term “bind” has been questioned during the years. First, the interaction was proposed to be a of receptor−ligand kind between the AFPs and the ice surface.14 Recently, new studies have pointed the attention to the ice/water interface. Wierzbicki et al.15 proposed that the protein is adsorbed onto the ice/water interface, arguing that a receptor−ligand type of interaction would not take into account the interface between the two water phases. The adsorption of AFPs onto the ice/ water interface was first proposed by Haymet et al.16 and then reported also in other studies.17,18 Many theoretical studies19−25 have been carried out on the ice/water interface and its properties, but to the best of our knowledge none of them include an AFP molecule. From the ice/water interface studies, it was reported that the four ice planes considered in those © XXXX American Chemical Society
studies show slightly different thickness of the interface. The different interfaces were estimated to be 10−20 Å thick22−25 going from the basal interface (the thickest) to the bipyramidal interface (the thinnest).23 AFPs show very different amino acid sequences and 3-D structures, suggesting a convergent evolution that gave different answers (in terms of protein structure) to the same problem of adaptability to an extreme environment. All AFPs share one common feature. They are all amphipathic having one hydrophilic and one hydrophobic surface, which probably is the origin of their thermal hysteresis activity. In terms of their activity, AFPs have been divided into two main categories: moderately active and hyperactive. In the work of Scotter et al.,26 different AFPs were grouped as follows: to the moderately active group belong, for example, AFPs from the winter flounder (wfAFP, isoform HPLC-6), the pure calcium-dependent type II of atlantic herring (ahAFP), and the shorthorn sculpin (ssAFP); while to the hyperactive group belong, for example, AFPs from the spruce budworm (sbwAFP), the Tenebrio molitor (TmAFP), the Antarctic bacterium Marinomonas primoryensis (MpAFP), and the snow flea (sfAFP). In this work, we have studied, through MD simulations, the influence of two AFPs on the ice/water interface of four different ice planes. We have studied one protein belonging to the hyperactive group (the sfAFP) and the most studied AFP belonging to the moderately active group (the wfAFP). Received: December 1, 2014 Revised: January 21, 2015
A
DOI: 10.1021/jp5119713 J. Phys. Chem. B XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry B
wfAFP), and PDB ID 1UBQ49 (ubiquitin). The potential energies of the both proteins were minimized in vacuum for 1000 steps with a Steepest Descent algorithm. The resulting structures were then used for the simulations at the ice/water interfaces. AFP/Ice/Liquid Water. The structures, obtained after the equilibration protocol followed for the ice/liquid water systems, were used to insert the protein in the two-phase water systems. The protein coordinates, obtained from the protein energy minimization in vacuum, were translated and rotated until the flat hydrophobic face of the protein was parallel to the ice plane. The starting configurations are presented in Figure 1. The starting box dimensions are equal to those of the ice/ liquid water systems. The empty volume needed for inserting the protein was obtained by removing all water molecules within 3 Å from the protein (∼450 molecules for sfAFP and ∼250 molecules for the wfAFP). The water potential energy was then minimized for 1000 steps with a Steepest Descent algorithm keeping the protein coordinates fixed. Thereafter, a 50 ps molecular dynamics simulation was performed by imposing constraints on all protein bonds. At this point, 10 ns MD simulations, without any constraints on the proteins atoms, were carried out for all systems at four different temperatures, 224, 225, 226, and 227 K. The trajectories were saved every 1 ps. Diffusion Coefficient and Other Properties. The diffusion coefficient profile was calculated for all four ice/ water interfaces with and without the AFPs bound to the ice surface. It was calculated for all water molecules within intervals 1 Å thick. The interval was then shifted by 0.5 Å along the longest box direction (depending on the simulated interface, see Table 1). The root-mean-square displacement of all molecules within all intervals was computed along the 10 ns trajectory, which was divided into 14 segments of 700 ps each. The diffusion coefficient D was calculated from the slope of the root-mean-square displacement of all water molecules within each interval and averaged over the entire simulation time. The value of the diffusion coefficient was estimated using the Einstein relation:50−52
Moreover, we have simulated a non-AFP protein as a control. We have estimated the ice/water interface thickness for systems with and without the AFPs adsorbed onto it at different temperatures. To estimate the interface thickness, we have computed the diffusion profile perpendicular to the ice plane direction for all four ice planes.
■
METHODS AND MODELS Force Field. All of the molecular dynamics simulations were carried out using Gromacs 4.5.427−30 simulation software. For the proteins the OPLS-AA force field31−37 was used, while water molecules were modeled with the TIP4P model.38 NPT ensemble was adopted for all simulations. Pressure was set to 1 atm and controlled by the Parrinello−Rahman barostat.39,40 Temperature was controlled by the Nose−Hoover thermostat.41,42 Cut-off radius for both Coulombic interactions (calculated using the PME method43,44) and Lennard-Jones interactions was 10 Å. All bonds involving hydrogen atoms were treated by the LINCS algorithm.45 Ice/Liquid Interface. The starting ice slabs were downloaded from the Sonnichsen lab.46 The starting ice slab was translated and replicated to create a simulation box with enough bulk water in both phases (liquid and solid). The final dimension of the boxes depends on the simulated ice plane. The box dimensions and the number of water molecules in both phases are shown in Table 1. Table 1. Box Dimensions and Number of Water Molecules ice plane prism secondary prism bipyramidal basal
box dimensions (Å)
water molecules (liquid phase)
water molecules (ice)
66 × 67 × 185 61 × 217 × 59
12 960 12 288
12 960 12 288
181 × 67 × 75 68 × 67 × 165
14 724 11 880
14 076 11 880
The simulation protocol is from our previous work.47 The potential energy of the initial configurations were minimized for 1000 steps with a Steepest Descent algorithm. Three short MD simulations (50 ps) were thereafter sequentially performed at increasing temperatures (100, 150, and 200 K) in an NPT ensemble imposing the position restraints on all water oxygen atoms. The system was thereafter simulated for 100 ps at 300 K allowing the water molecules in one-half of the box to move freely. In this way, we obtained all four ice planes at the interface with water. The final configuration was used as the starting structure for the ice/liquid water simulations. For all four ice planes, a 10 ns MD was performed in an NPT ensemble storing the configurations every 1 ps. Five different temperatures were chosen for the trajectory production: 224, 225, 226, 227, and 228 K. The temperatures range from the melting point of hexagonal ice (Ih) obtained in our previous study47 to temperature 4 K below that point. Proteins Energy Minimization. The starting coordinates of the sfAFP were obtained from the crystal structure solved by X-ray diffraction at 0.98 Å resolution. The starting coordinates of the type I AFP were obtained from the crystal structure solved by X-ray diffraction at 1.50 Å resolution. The starting coordinates of the ubiquitin were obtained from the crystal structure solved by X-ray diffraction at 1.80 Å resolution. The three sets of coordinates were obtained from the Protein Data Bank,48 PDB ID 2PNE5 (sfAFP), PDB ID 1WFB6 (type I AFP,
D = lim
t →∞
⟨[r(t ) − r(0)]2 ⟩ 6kT
The solvent accessible surface (SAS) is computed using GROMACS tool g_sas. Atoms are classified as hydrophilic and hydrophobic according to their partial charge. Hydrophobic atoms have a partial charge between −0.2 and +0.2. Hydrogen bonds (HB) are defined by a donor−acceptor cutoff distance of 3.4 Å and a ∠DHA angle of 45°.
■
RESULTS AND DISCUSSION Ice/Liquid Interface. The evolution of each system was followed by the total energy profile. For the TIP4P water model, a melting temperature (Tm) of 228−229 K was estimated by previous studies.47,53 Following the drift of the energy profile (see Figure 1), we found a Tm of 228 K for the secondary prism and bipyramidal ice plane and between 227 and 228 K for the prism and the basal plane. The Tm is evaluated by the drift change of the energy profile.53 The Tm is obtained at the temperature when the drift passes from positive (ice is melting) to negative (ice is growing). The obtained results are satisfactory and within an error of ±2 K. B
DOI: 10.1021/jp5119713 J. Phys. Chem. B XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry B
Figure 1. Starting configuration of the protein/ice/water systems: (a) sfAFP; (b) wfAFP; (c) ubiquitin. In all systems, water (liquid) is in red, ice is in cyan, and the protein is in green.
order of the interfaces and their widths are found in good agreement with previous estimations of ice/water interface width of ∼10−20 Å.22−25 Also, the interface span (2 Å) is consistent with the estimate of Hayward and Haymet.23,24 AFPs at the Ice/Liquid Interface. The same procedure for the estimation of the ice/water interface thickness was applied also to the systems containing the proteins. The interface widths for all ice planes are presented in Table 3. It can be easily seen that the interface thickness has become wider for all ice planes for the two AFPs. To evaluate the influence of the two AFPs, we have reported the width difference as a percentage, calculated with respect to the values of the ice/ water systems (see Table 2). The two AFPs affect the interface width to different extents. The sfAFP leads to an increase of the interface width of ∼25−40%, while for the wfAFP the difference is estimated to be about 10−20%. The sfAFP shows a width increase very similar for all ice planes except for the basal plane, which is slightly less broadened. The wfAFP instead induces an increase of the interface for the secondary
All simulation boxes were built to contain enough bulk water molecules. The box axis containing the ice/water interface was set to be the longest box dimension. As shown in Figure 2, the periodic boundary conditions create two ice/water interfaces for each box. In this way, we could get twice as much diffusion data for each ice plane. The thickness of the interface was estimated from the diffusion profiles (see Supporting Information Figure S1) through the 10−90% width.19−21,23,24 The 10−90% width can be used to estimate the thickness of any interface. It is defined as the length over which a certain property varies from 10% to 90% of the bulk value (liquid water in our case) when passing across the interface from the solid to the liquid phase. The widths of the different ice planes are presented in Table 2. The values are obtained as averages over the five different temperatures set in the simulations for the two interfaces of each box. The bulk water values were calculated from the bulk water region of each box (Figure 3). The estimated widths going from the thickest to the thinnest interface are basal, prism, bipyramidal, secondary prism. The C
DOI: 10.1021/jp5119713 J. Phys. Chem. B XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry B
Figure 2. Evolution of the total energy for the four ice planes at different temperatures.
are consistent with the classification of the two AFPs given by Scotter et al.26 They assign the sfAFP to the hyperactive family, while the wfAFP belong to the moderately active group. SfAFP has a thermal hysteresis activity of 1.0 °C at a concentration of 0.02 mM, while the wfAFP produces
