Subscriber access provided by NEW YORK UNIV
Article
Deciphering #-lactoglobulin interactions at an oilwater interface: a molecular dynamics study Davoud Zare, Kathryn M McGrath, and Jane Rosemary Allison Biomacromolecules, Just Accepted Manuscript • DOI: 10.1021/acs.biomac.5b00467 • Publication Date (Web): 19 May 2015 Downloaded from http://pubs.acs.org on May 25, 2015
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Biomacromolecules is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
Deciphering β-lactoglobulin interactions at an oilwater interface: a molecular dynamics study Davoud Zare¶,†, Kathryn M. McGrath¶,†, Jane R. Allison‡,§,#,* ¶. MacDiarmid Institute for Advanced Materials and Nanotechnology, School of Chemical and Physical Sciences, Victoria University of Wellington, PO Box 600, Wellington 6140, New Zealand †. Riddet Institute, Massey University, Private Bag 11 222, Palmerston North 4442, New Zealand ‡. Centre for Theoretical Chemistry and Physics, Institute of Natural and Mathematical Sciences, Massey University Auckland, Albany, Auckland 0632, New Zealand §. Biomolecular Interaction Centre, University of Canterbury, Private Bag 4800, Christchurch 8140, New Zealand # Maurice Wilkins Centre for Molecular Biodiscovery, University of Auckland, Private Bag 92019, Auckland, New Zealand
β-lactoglobulin, protein adsorption, molecular dynamics, GROMACS, GROMOS
ACS Paragon Plus Environment
1
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 2 of 22
Protein adsorption at liquid-liquid interfaces is of immense relevance to many biological processes and dairy-based functional foods. Due to experimental limitations, however, there is still a remarkable lack of understanding of the adsorption mechanism, particularly at a molecular level. In this study, atomistic molecular dynamics simulations were used to elucidate the approach and adsorption mechanism of β-lactoglobulin (β-LG) at a decane-water interface. Through multiple independent simulations starting from three representative initial orientations of β-LG relative to the decane surface the rate at which β-LG approaches the oil/water interface is found to be independent of its initial orientation, and largely stochastic in nature. While the residues that first make contact with the decane and the final orientation of β-LG upon adsorption are similar in all cases, the adsorption process is driven predominantly by structural rearrangements that preserve the secondary structure but expose hydrophobic residues to the decane surface. This detailed characterisation of the adsorption of β-LG at an oil/water interface should inform the design and development of novel encapsulation and delivery systems in the food and pharmaceutical sciences.
Introduction Protein adsorption at an interface is integral to a broad range of dairy-based functional foods, However, full elucidation of detailed information at a high-resolution (atomic level) on the conformational changes of proteins adsorbed at, for example, an oil/water (O/W) interface remains challenging, despite constant improvement over the past decades of experimental techniques such as nuclear magnetic resonance (NMR) spectroscopy, electron microscopy, X-ray diffraction, Fourier transform infrared spectroscopy (FTIR), and circular dichroism (CD), which are used to determine the structures of proteins1.
ACS Paragon Plus Environment
2
Page 3 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
Milk proteins are among the most common proteins used in the food industry due to their excellent natural amphiphilicity, abundance, low cost and nutritional importance. The two main classes of milk proteins are caseins and whey proteins. Whey proteins have compact globular structures and can be fractionated into β-lactoglobulin (β-LG), bovine serum albumin, αlactalbumin and immunoglobulin. β-LG is the predominant component, and has a major role in the processing of whey into functional products. This 162 amino acid protein (~18 kDa) contains two intra-molecular disulfide linkages and a single free hidden thiol group2, 3. It has well-defined secondary and tertiary structure, forming a β-barrel fold with a large internal cavity that binds lipids and related molecules4-7. Depending on pH, temperature, and salt concentration, β-LG can exist as a monomer, dimer, or octamer8-10. Most physicochemical properties of O/W food emulsions stabilized by milk proteins, such as texture, viscosity, stability and digestibility, are largely determined by the structural properties of the adsorbed layer at the O/W interface11-13. Because of the importance and usage of milk proteins, there have been some experimental studies on the conformational behaviour of β-LG at O/W interfaces using FTIR14-17 and CD15, 18-21, but their results are contradictory. Fang and Dalgleish14 investigated conformational changes of β-LG upon adsorption to the O/W interface for 20% (w/w) soybean oil emulsions at different pH values (6 and 7) and protein concentrations (1% and 2% w/w). They observed changes in the secondary structure upon adsorption; specifically, decreased intra-molecular β-sheet content, unchanged α-helix content, increased unordered structure and increased inter-molecular β-sheet formation. The pattern of structural changes obtained with FTIR was very similar for both protein concentrations except that at the lower protein concentration, the conformational rearrangements were quicker and more extensive so as to better cover the hydrophobic O/W interface and form a stable emulsion. Lower
ACS Paragon Plus Environment
3
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 4 of 22
pH also resulted in faster and more extensive conformational changes. Additionally, the adsorbed protein at the O/W interface at pH 6 had a greater proportion of unordered structure and (possibly) inter-molecular β-sheet whereas at pH 7 it had higher α-helical content14. CD has been used to study protein adsorption to flat solid interfaces, colloidal silica and latex beads, but its use to study protein adsorption at O/W interfaces is complicated by the significant amount of scattered light produced by small emulsion droplets (100 nm-10 µm). To overcome this problem and allow CD to be performed on emulsions Husband et al.15 developed a refractive index matched emulsion (RIME) method, in which the refractive index (RI) of the continuous phase is matched to that of the oil phase by adding glycerol and polyethylene glycol. In contrast to the FTIR measurements, RIME CD appeared to indicate an increase in α-helix content upon βLG adsorption to emulsion interfaces. These differences may be attributable to the aforementioned difficulty in obtaining CD spectra of proteins in emulsions. It is also possible that adding glycerol or polyethylene glycol may affect the protein conformation. Recently, however, synchrotron radiation CD confirmed the increase in α-helical content of β-LG upon adsorption to emulsion O/W interfaces18-20. Advanced instrumental methods allowing direct access to high resolution molecular information, especially changes in tertiary structure, are limited. Furthermore, for complex experimental systems, such as polydisperse emulsions, many different interfaces exist, with the protein responding to each local environment in a different way. One alternative, nonexperimental method that provides an opportunity to study the details of the protein conformational changes upon interaction with an O/W interface is molecular dynamics (MD) simulation. With the advancements of high performance computing, MD simulation serves as a
ACS Paragon Plus Environment
4
Page 5 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
powerful tool providing complementary information to experiments and allowing elucidation of the changes to protein secondary and tertiary structure stimulated by adsorption at an interface. Most simulation studies conducted to date have been confined to studies dealing with protein adsorption at liquid-solid interfaces21-24, protein adsorption on different self-assembled monolayers25, 26 or protein denaturation upon heating or addition of salts in solution27, 28. Recently, however, the interfacial and adsorption behaviour of proteins, surfactants and nanoparticles at different liquid-liquid interfaces has been studied using MD29-32. Both atomistic and coarse-grained simulations of barley lipid transfer protein adsorption at a decane-water interface showed the lipid transfer protein adsorbing with the helical region parallel to the interface. The average tilt angle normal to the interface was 73° for the all-atom model and 62° for the coarse-grained model31. The atomistic simulation showed the secondary structure of the protein to be conserved upon adsorption. MD simulations of nanoparticles (NPs) and nonionic surfactants at an O/W interface showed that at low surfactant concentration, NPs and surfactants have a synergistic effect in lowering the interfacial tension32. Adsorption of surfactant at the NP surface prevents NP aggregation, and also reduces the surfactant efficiency in lowering the surface tension between oil and water. A recent MD investigation of hydrophobin adsorption at an octane-water interface found adsorption to be essentially irreversible, and showed that surface structure and the flexibility of the protein have a large influence on the interfacial adsorption strength29. This study probed the adsorption of β-LG at an O/W (decane/water) interface using atomiclevel MD simulations, allowing the mechanism of adsorption to a purely hydrophobic surface and concomitant structural changes to the protein to be followed in detail. The ability of β-LG to
ACS Paragon Plus Environment
5
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 6 of 22
approach and interact with the decane surface, including its orientation with respect to the surface upon adsorption and the residues involved in the adsorption process were independent of its initial orientation with respect to the surface. The parts of β-LG that contribute to its final, stable adsorption on the decane surface were almost identical in all cases, however, driven by favourable van der Waals interactions between β-LG and decane. These interactions involve hydrophobic residues from the N-terminal region and the interior of the calyx, which become exposed due to significant rearrangement of the tertiary structure, although the secondary structure remains remarkably intact both during and after adsorption.
Methods Simulations All simulations were prepared and performed with the GROMACS simulation package version 4.5.433 and GROMOS 43A2 force field unless otherwise stated. Coulombic and van der Waals interactions were computed exactly within a cut-off distance of 1.0 nm. Outside of this, the longrange electrostatic interactions were treated using the particle mesh Ewald method34, 35 with a maximal spacing for the FFT grid of 0.16 nm and an interpolation order of 4. Coordinates and the topology file for a single decane molecule were obtained from S. Euston 31
. A slab comprising 818 decane molecules was built by random insertion of decane molecules
followed by energy minimisation, a 100 ps NVT simulation during which the temperature was increased from 50 K to 300 K, and finally a semi-isotropic 240 ps NpT simulation in which the area of the xy-surface was kept constant and the z-dimension allowed to vary until the experimental density of decane [720 g/L] was obtained. The final xyz dimensions of the oil slab were 8.56 × 8.56 × 3.43 nm, sufficient for the subsequent insertion and simulation of β-LG with
ACS Paragon Plus Environment
6
Page 7 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
cutoff distances of 1.0 nm and periodic boundary conditions. Prior to inserting β-LG, the dimensions in the z direction were increased to 12 nm so that the portion of the box into which the protein was inserted was the same size in all dimensions. Initial coordinates for β-LG, including crystal water molecules, were obtained from the X-ray structure (PDB ID: 3BLG). β-LG was added to the part of the system not occupied by the decane slab in three different orientations: rot0 (unmodified PDB coordinates, three replicates denoted rot0-1 etc); rot90x (rotated 90° around the x-axis, two replicates); rot90y (rotated 90° around the y-axis, two replicates) (Figure 1B). In all cases, the initial distance of the centre of mass of β-LG from the decane surface in the z-direction dinit was approximately 4 nm (Figure 1A). The simulation box was filled with SPC water molecules36 and the entire system subjected to energy minimization, heated from 50 K to 300 K over 100 ps in the NVT ensemble, and equilibrated for a further 1 ns in the NpT ensemble. The temperature and the pressure were maintained using the Berendsen thermostat and barostat37, with coupling times of τT = 0.1 ps and τp = 2 ps, respectively. The lengths of all bonds involving hydrogen atoms were constrained using LINCS38 with an order of 4, allowing for an integration time step of 2 fs. Each system was subsequently simulated for 200 ns, with coordinates saved every 5 ps. Analysis The majority of the analysis of the simulations was carried out using GROMACS analysis tools unless otherwise stated. The atom positional root mean square deviation (RMSD) from the initial crystallographic structure was calculated for the Cα atoms only after superimposing the Cα atoms onto their initial positions. The radius of gyration (Rg) was also calculated for the Cα atoms only. Secondary structure was calculated using the GROMACS do_dssp program, which
ACS Paragon Plus Environment
7
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 8 of 22
implements the algorithm of Kabsch and Sander39, and post-processed prior to plotting using a self-made Python script. The orientation angle θ was defined as the angle between the unit vector normal (n) to the decane surface (approximated by the z-axis) and the unit vector between the Cα atoms of residues 11 and 34, which approximates the long axis of β-LG in its natively folded structure (Figure 1A). The residues that come into contact with the oil slab were found using a self-made Tcl script in VMD40 to identify residues for which at least one atom was within dcontact = 0.6 nm of any atom of a decane molecule (Figure 1). The minimum distance between the protein and the decane slab was calculated using the GROMACS g_mindist program. The change in the hydrophobic surface area of β-LG was calculated using the GROMACS g_sas program without taking contact with decane into account, i.e. including the contribution from atoms that would be covered by the decane and so not solvent accessible.
Results and Discussion Approach of β-LG to the decane surface While in all cases, β-LG transports from solution (middle of the simulation box) to the oil surface, to which it eventually adsorbs, the time of approach varies both between replicates of a particular orientation of the protein with respect to the surface and between different orientations (Figure 2). For the three rot0 systems, β-LG approaches the decane surface very rapidly, within the first 20-30 ns, whereas the two rot90x replicates approach at times ranging from 60-100 ns (Figure 2A) Interestingly, rot90y-2 is among the first to approach the decane surface, whereas rot90y-1 is the last. Therefore the rate at which β-LG approaches the O/W interface is independent of its initial orientation, and is largely stochastic in nature.
ACS Paragon Plus Environment
8
Page 9 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
The initial contact between β-LG and the decane surface does not necessarily result in lasting contact formation and adsorption. Rot0-1 adheres to the surface quickly, with the number of protein residues in contact with the decane rapidly increasing to around 70 upon β-LG first contacting the surface, whereas rot90x-2 and rot90y-1, for instance, make multiple approaches to the surface before finally adsorbing at around 100 and 130 ns, respectively (Figure 2). Merely being close to the oil surface is therefore insufficient for adsorption to occur. To quantitatively characterize the approach of β-LG to the decane surface, the protein orientation, as defined in Figure 1A, was calculated throughout each simulation. During the first phase, when the β-LG is free in solution, the orientation angle θ undergoes significant fluctuation, ranging from ~60° in the case of rot90y-2 to almost 180° for rot90y-1 (Figure 2B), but the ultimate θ values once adsorption has occurred are in the range 90-135° in all cases. This is a surprisingly narrow range given the degree of deformation of the tertiary structure of β-LG upon adsorption (see Figure 4), indicating that the approach of the protein to the decane surface is not constrained by its initial orientation. In each simulation, the plateau in the value of θ coincides with β-LG contacting the decane surface (dcontact < 0.6 nm). The protein does not absorb immediately to the decane surface, but diffuses on the surface, as evidenced by the fluctuation in the x and y components and the overall magnitude |z| of the protein–decane centre of mass distance (Supporting Information Figures 1 and 2), but its orientation with respect to the surface does not change markedly during the adsorption process. Process of adsorption of β-LG to the decane surface
ACS Paragon Plus Environment
9
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 10 of 22
The convergence towards similar orientations of β-LG relative to the decane surface upon stable adsorption regardless of the initial orientation suggests that the same residues are involved in adsorption in all cases. To investigate whether successful adsorption requires particular residues of β-LG to come into contact with the oil surface, and where the adsorbing residues are located in protein structure, we identified the residues that are closest to the decane surface throughout the process of adsorption (Figure 3 and Supporting Information Figure 3). As discussed above, β-LG does not absorb instantly upon approaching the decane surface, but rather makes multiple approaches to the surface and diffuses laterally on the surface before finally adsorbing. During these initial phases, the number and identity of the residues in contact with the decane varies through time and between replicates (Figure 3 and Supporting Information Figure 3). Residues from the N-terminal region and around positions 50, 75, 100 (excluding the rot0 replicates), 110 (excluding rot90y-2) and 130-140 are generally the first to approach the decane. Within these regions there are specific residues, such as Lys77 and Lys14, which form contacts in almost all simulations. Later in the adsorption process, additional groups of residues around positions 16-35, 70-85 and 130-150 also come into contact with the decane, although again the exact residues involved differ between simulations. These later contacts are largely retained and thus can be considered to be drivers of stable adsorption. Notably, there are a large number of hydrophobic residues in the N-terminal region (Leu1, Ile2, Val3, Met7, Leu10, Ile12, Val15, Ala16, Leu22, Ala23, Met24, Ala25, Ala26, Ile29), and residues 70-123 correspond to the hydrophobic residues that line the calyx (β-strands E, F, D, G, Figure 3).
ACS Paragon Plus Environment
10
Page 11 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
Ultimately, ~80-90 residues are in contact with the decane surface once adsorption has taken place – approximately half of this 162-residue protein. Key regions that do not typically bind to the decane surface are the C-terminus (other than rot90y-2) and, in some but not all replicates, residues around 30-50, 60-70 and 120-130, with slight differences between the replicates in the exact location of the non-binding regions. These results indicate that while the initial contact between β-LG and the decane surface typically involves particular residues, e.g. Lys77 and Lys14, stable adsorption of β-LG to the decane surface is non-specific, driven by the predominantly hydrophobic N-terminal region and by rearrangement of the structure (Figure 5) to expose the hydrophobic residues that line the lipid-binding cavity. This is confirmed by the increase in the hydrophobic surface area (SASA) of the protein that occurs during the adsorption process (Figure 2E). The interactions between the newly exposed parts of the protein and the decane surface are energetically favourable, with the evolution of the protein-decane interaction energy closely tracking the approach of the protein to the surface (Figure 2). The strong correlation between the number of protein-decane contacts formed and the protein-decane interaction energy further indicates that the adsorption energy is not dominated by contributions from a few residues. Structural changes of β-LG upon adsorption The structural integrity of β-LG was monitored by calculating the atom-positional rmsd of the Cα atoms with respect to the initial structure of each simulation and the Rg (Figure 4A and B). The rmsd of a simulation of β-LG in solution without a decane slab present was also calculated for comparison. In all simulations with a decane slab present, the rmsd increases to be at least twice that of β-LG alone. The time at which the rmsd increases, however, differs between
ACS Paragon Plus Environment
11
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 12 of 22
simulations; in general, the rmsd increases somewhat later than the initial contact between β-LG and the decane surface occurs (Figure 2). The Rg values follow a similar trend to the rmsd, increasing some time after initial contact between β-LG and the decane surface. The nature of the changes in the overall shape of β-LG upon adsorption to the decane surface (Figure 5) is revealed in the decomposition of the Rg (Supporting Information Figure 4). The z component Rg,z in general tracks the overall Rg and increases throughout each simulation, whereas the x and y components are generally of lower magnitude. Somewhat counterintuitively, therefore, β-LG actually stretches in the z direction, i.e. normal to the decane surface, during the later stages of the adsorption process. Together, these results point to the structure of β-LG being affected little by the initial contact to the decane surface, and only undergoing significant change later, at the point of adsorption to the decane surface. The secondary structure content was also monitored (Figure 4C and D and Supporting Information Figure 5). While the tertiary structure of β-LG changes considerably upon its adsorption at the decane surface, the secondary structure is well conserved throughout these structural changes, with close to 100% of the protein retaining its initial classification into a given type of regular structure (Figure 4C and D, Figure 5 and Supporting Information Figure 5). Adsorption to the surface therefore results in rearrangement of secondary structure elements, rather than a complete loss of structure. This is in contrast to the results of Fang and Dalgleish14, who observed a decrease in intra-molecular β-sheet content upon adsorption. This was accompanied by a concomitant increase in inter-molecular β-sheet content, something that cannot be investigated in single molecule simulations. The increase in α-helix observed with
ACS Paragon Plus Environment
12
Page 13 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
RIME CD15 and synchrotron radiation19, 20 is also not observed here, most likely due to the limited time scale of the simulations compared to the experimental time scale. While coarsegrained simulations would allow longer time scales to be accessed, they are inappropriate for probing structural changes due to the common requirement for secondary structure constraints. It will be of interest, therefore, to see whether additional experimental studies can resolve the current discrepancy between the secondary structure changes observed by RIME CD and FTIR. Matched computational and experimental studies, such as using steered MD simulations to mimic the single-molecule force spectroscopy that has been carried out for β-LG at an O/W interface42, also hold much promise. Conclusions The adsorption of proteins to O/W interfaces occurs in many biological processes and is critical to the physicochemical properties of functional foods, particularly those based on emulsions stabilised by milk proteins. While low-resolution structural properties of adsorbed proteins can be measured experimentally, detailed characterisation of the process of adsorption and concomitant structural changes requires an alternative approach. Here, MD simulations of βLG in the presence of a decane slab allowed the approach and adsorption of β-LG to the O/W interface to be followed in atomic-level detail. The approach phase of the adsorption was largely stochastic in nature. The residues that make contact with the decane surface and the final orientation of β-LG with respect to the surface are independent of the initial orientation of β-LG. The adsorption process was found to be largely driven by groups of predominantly hydrophobic residues, some of which become exposed during adsorption by structural rearrangements. These structural changes preserve the secondary structure of β-LG, which is in contrast to some experimental results; these discrepancies are most likely due to differences between the nature
ACS Paragon Plus Environment
13
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 14 of 22
and time-scale of the experimental and computational studies. The detailed characterisation of the adsorption of β-LG at an O/W interface presented here, particularly the identification of the residues that drive the adsorption process and their location in the native structure of β-LG will enable improved engineering of functional food products and drug delivery systems.
Supporting Information. β-LG-decane distance, β-LG adsorption landscapes, distance to decane of each residue during adsorption, Rg of β-LG, secondary structure content of β-LG. This material is available free of charge via the Internet at http://pubs.acs.org.
ACS Paragon Plus Environment
14
Page 15 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
Figure 1. Schematic diagrams of the simulation system. A) Definition of the protein-surface orientation angle θ and the protein-surface separation distances dinit and dcontact. B) An example of the rotation procedure used to generate the different initial orientations.
ACS Paragon Plus Environment
15
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 16 of 22
Figure 2. Approach of β-LG to the decane surface. Time-series of A) minimum distance between β-LG and decane surface; B) orientation angle θ between β-LG and the decane surface; C) number of β-LG residues in contact with decane surface; D) interaction energy between β-LG and decane and E) hydrophobic surface area. The protein-decane potential energy is used as a
ACS Paragon Plus Environment
16
Page 17 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
proxy for the adsorption energy; only the van der Waals energy contributes, as the electrostatic component is zero due to the decane molecules having zero partial charge.
Figure 3. Residues involved in β-LG adsorption. Contact between a β-LG residue and the decane surface (dcontact < 0.6 nm) is indicated in blue (hydrophilic) or red (hydrophobic). The eight β-strands and two α-helices of β-LG are labelled on the structure shown in the top right panel and their location in the sequence indicated by the vertical panels.
ACS Paragon Plus Environment
17
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 18 of 22
Figure 4. Structural properties of β-LG. Time-series of the A) atom-positional rmsd of the Cα atoms with respect to the initial structure; B) Rg; C) percentage of residues exhibiting each type of secondary structure; and D) secondary structure of each residue during the simulations. Panels C) and D) show data for the rot0-1 simulation only; the secondary structure content of β-LG in the other simulations is similarly constant throughout each simulation (Supporting Information Figure 5).
ACS Paragon Plus Environment
18
Page 19 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
Figure 5. Structural changes to β-LG upon adsorption. Snapshots of β-LG at the start of each simulation (top) as labelled, and at the end, after adsorption to the decane surface (bottom).
ACS Paragon Plus Environment
19
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 20 of 22
Corresponding Author * Correspondence should be addressed to Jane R. Allison, [email protected]. Author Contributions The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Funding Sources JRA received financial support from a Massey University Research Fund Early Career Grant and a Marsden Fund Fast Start award (13-MAU-039).
The authors wish to acknowledge Bluefern (University of Canterbury) for provision of supercomputing facilities, and Stephen R. Euston (Herriot-Watt University) for providing coordinate and topology files for decane.
β-LG, β-lactoglobulin; O/W, oil/water; Rg, radius of gyration; rmsd, root-mean-square deviation.
ACS Paragon Plus Environment
20
Page 21 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
References (1) Zhai, J. l.; Day, L.; Aguilar, M.-I.; Wooster, T. J. Curr. Opin. Colloid Interface Sci. 2013, 18, 257. (2) Singh, H. Food Hydrocolloids 2011, 25, 1938. (3) Thompson, A.; Boland, M.; Singh, H.; (eds) Milk Proteins: from Expression to Food; 2nd ed.; Academic Press, Elsevier Inc., 2009. (4) Brownlow, S.; Cabral, J. H. M.; Cooper, R.; Flower, D. R.; Yewdall, S. J.; Polikarpov, I.; North, A. C. T.; Sawyer, L. Structure 1997, 5, 481. (5) Kontopidis, G.; Holt, C.; Sawyer, L. J. Mol. Biol. 2002, 318, 1043. (6) Qin, B. Y.; Bewley, M. C.; Creamer, L. K.; Baker, H. M.; Baker, E. N.; Jameson, G. B. Biochemistry 1998, 37, 14014. (7) Uhrínová, S.; Smith, M. H.; Jameson, G. B.; Uhrín, D.; Sawyer, L.; Barlow, P. N. Biochemistry 2000, 39, 3565. (8) Kumosinski, T. F.; Timasheff, S. N. J. Am. Chem. Soc. 1966, 88, 5635. (9) Mckenzie, H. A., Sawyer, W. H Nature 1967, 214, 1101. (10) Sakurai, K.; Oobatake, M.; Goto, Y. Prot. Sci. 2001, 10, 2325. (11) Dalgleish, D. G. Food Hydrocolloids 2006, 20, 415. (12) Kim, H. J.; Decker, E. A.; McClements, D. J. J Agric Food Chem 2002, 50, 7131. (13) Kim, H. J.; Decker, E. A.; McClements, D. J. Langmuir 2002, 18, 7577. (14) Fang, Y.; Dalgleish, D. G. J. Colloid Interface Sci. 1997, 196, 292. (15) Husband, F. A.; Garrood, M. J.; Mackie, A. R.; Burnett, G. R.; Wilde, P. J. J Agric Food Chem 2001, 49, 859. (16) Lee, S.-H.; Lefèvre, T.; Subirade, M.; Paquin, P. J Agric Food Chem 2007, 55, 10924. (17) Sakuno, M. M.; Matsumoto, S.; Kawai, S.; Taihei, K.; Matsumura, Y. Langmuir 2008, 24, 11483. (18) Day, L.; Zhai, J.; Xu, M.; Jones, N. C.; Hoffmann, S. V.; Wooster, T. J. Food Hydrocolloids 2014, 34, 78. (19) Zhai, J.; Miles, A. J.; Pattenden, L. K.; Lee, T.-H.; Augustin, M. A.; Wallace, B. A.; Aguilar, M.I.; Wooster, T. J. Biomacromolecules 2010, 11, 2136. (20) Zhai, J.; Wooster, T. J. Langmuir 2011, 27, 9227. (21) Zhao, D.; Peng, C.; Zhou, J. Phys. Chem. Chem. Phys. 2015, 17, 840. (22) Hagiwara, T.; Sakiyama, T.; Watanabe, H. Langmuir 2009, 25, 226. (23) Wei, T.; Carignano, M. A.; Szleifer, I. Langmuir 2011, 27, 12074. (24) Wei, T.; Carignano, M. A.; Szleifer, I. J. Phys. Chem. B 2012, 116, 10189. (25) Liu, J.; Liao, C.; Zhou, J. Langmuir 2013, 29, 11366. (26) Peng, C.; Liu, J.; Zhao, D.; Zhou, J. Langmuir 2014, 30, 11401. (27) Euston, S. R. Food Hydrocolloids 2013, 30, 519. (28) Euston, S. R.; Ur-Rehman, S.; Costello, G. Food Hydrocolloids 2007, 21, 1081. (29) Cheung, D. L. Langmuir 2012, 28, 8730. (30) Cheung, D. L. Adv. Chem. 2014, 2014, 12. (31) Euston, S. R. Biomacromolecules 2010, 11, 2781. (32) Ranatunga, R. J. K. U.; Nguyen, C. T.; Wilson, B. A.; Shinoda, W.; Nielsen, S. O. Soft Matter 2011, 7, 6942. (33) Hess, B.; Kutzner, C.; van der Spoel, D.; Lindahl, E. J. Chem. Theory Comput. 2008, 4, 435. (34) Darden, T.; York, D.; Pedersen, L. J. Chem. Phys. 1993, 98, 10089. (35) Essmann, U.; Perera, L.; Berkowitz, M. L.; Darden, T.; Lee, H.; Pedersen, L. G. J. Chem. Phys. 1995, 103, 8577. (36) Berweger, C. D.; van Gunsteren, W. F.; Müller-Plathe, F. Chem. Phys. Lett. 1995, 232, 429. (37) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; DiNola, A.; Haak, J. R. J. Chem. Phys. 1984, 81, 3684.
ACS Paragon Plus Environment
21
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 22 of 22
(38) Hess, B.; Bekker, H.; Berendsen, H. J. C.; Fraaije, J. G. E. M. J. Comput. Chem. 1997, 18, 1463. (39) Kabsch, W.; Sander, C. Biopolymers 1983, 22, 2577. (40) Humphrey, W.; Dalke, A.; Schulten, K. J. Mol. Graph. 1996, 14, 33. (41) Pieranski, P. Physical Review Letters 1980, 45, 569. (42) Touhami, A.; Alexander, M.; Kurylowicz, M.; Gram, C.; Corredig, M.; Dutcher, J. R. Soft Matter 2011, 7, 10274.
TOC Graphic
ACS Paragon Plus Environment
22
Article
Deciphering #-lactoglobulin interactions at an oilwater interface: a molecular dynamics study Davoud Zare, Kathryn M McGrath, and Jane Rosemary Allison Biomacromolecules, Just Accepted Manuscript • DOI: 10.1021/acs.biomac.5b00467 • Publication Date (Web): 19 May 2015 Downloaded from http://pubs.acs.org on May 25, 2015
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Biomacromolecules is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
Deciphering β-lactoglobulin interactions at an oilwater interface: a molecular dynamics study Davoud Zare¶,†, Kathryn M. McGrath¶,†, Jane R. Allison‡,§,#,* ¶. MacDiarmid Institute for Advanced Materials and Nanotechnology, School of Chemical and Physical Sciences, Victoria University of Wellington, PO Box 600, Wellington 6140, New Zealand †. Riddet Institute, Massey University, Private Bag 11 222, Palmerston North 4442, New Zealand ‡. Centre for Theoretical Chemistry and Physics, Institute of Natural and Mathematical Sciences, Massey University Auckland, Albany, Auckland 0632, New Zealand §. Biomolecular Interaction Centre, University of Canterbury, Private Bag 4800, Christchurch 8140, New Zealand # Maurice Wilkins Centre for Molecular Biodiscovery, University of Auckland, Private Bag 92019, Auckland, New Zealand
β-lactoglobulin, protein adsorption, molecular dynamics, GROMACS, GROMOS
ACS Paragon Plus Environment
1
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 2 of 22
Protein adsorption at liquid-liquid interfaces is of immense relevance to many biological processes and dairy-based functional foods. Due to experimental limitations, however, there is still a remarkable lack of understanding of the adsorption mechanism, particularly at a molecular level. In this study, atomistic molecular dynamics simulations were used to elucidate the approach and adsorption mechanism of β-lactoglobulin (β-LG) at a decane-water interface. Through multiple independent simulations starting from three representative initial orientations of β-LG relative to the decane surface the rate at which β-LG approaches the oil/water interface is found to be independent of its initial orientation, and largely stochastic in nature. While the residues that first make contact with the decane and the final orientation of β-LG upon adsorption are similar in all cases, the adsorption process is driven predominantly by structural rearrangements that preserve the secondary structure but expose hydrophobic residues to the decane surface. This detailed characterisation of the adsorption of β-LG at an oil/water interface should inform the design and development of novel encapsulation and delivery systems in the food and pharmaceutical sciences.
Introduction Protein adsorption at an interface is integral to a broad range of dairy-based functional foods, However, full elucidation of detailed information at a high-resolution (atomic level) on the conformational changes of proteins adsorbed at, for example, an oil/water (O/W) interface remains challenging, despite constant improvement over the past decades of experimental techniques such as nuclear magnetic resonance (NMR) spectroscopy, electron microscopy, X-ray diffraction, Fourier transform infrared spectroscopy (FTIR), and circular dichroism (CD), which are used to determine the structures of proteins1.
ACS Paragon Plus Environment
2
Page 3 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
Milk proteins are among the most common proteins used in the food industry due to their excellent natural amphiphilicity, abundance, low cost and nutritional importance. The two main classes of milk proteins are caseins and whey proteins. Whey proteins have compact globular structures and can be fractionated into β-lactoglobulin (β-LG), bovine serum albumin, αlactalbumin and immunoglobulin. β-LG is the predominant component, and has a major role in the processing of whey into functional products. This 162 amino acid protein (~18 kDa) contains two intra-molecular disulfide linkages and a single free hidden thiol group2, 3. It has well-defined secondary and tertiary structure, forming a β-barrel fold with a large internal cavity that binds lipids and related molecules4-7. Depending on pH, temperature, and salt concentration, β-LG can exist as a monomer, dimer, or octamer8-10. Most physicochemical properties of O/W food emulsions stabilized by milk proteins, such as texture, viscosity, stability and digestibility, are largely determined by the structural properties of the adsorbed layer at the O/W interface11-13. Because of the importance and usage of milk proteins, there have been some experimental studies on the conformational behaviour of β-LG at O/W interfaces using FTIR14-17 and CD15, 18-21, but their results are contradictory. Fang and Dalgleish14 investigated conformational changes of β-LG upon adsorption to the O/W interface for 20% (w/w) soybean oil emulsions at different pH values (6 and 7) and protein concentrations (1% and 2% w/w). They observed changes in the secondary structure upon adsorption; specifically, decreased intra-molecular β-sheet content, unchanged α-helix content, increased unordered structure and increased inter-molecular β-sheet formation. The pattern of structural changes obtained with FTIR was very similar for both protein concentrations except that at the lower protein concentration, the conformational rearrangements were quicker and more extensive so as to better cover the hydrophobic O/W interface and form a stable emulsion. Lower
ACS Paragon Plus Environment
3
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 4 of 22
pH also resulted in faster and more extensive conformational changes. Additionally, the adsorbed protein at the O/W interface at pH 6 had a greater proportion of unordered structure and (possibly) inter-molecular β-sheet whereas at pH 7 it had higher α-helical content14. CD has been used to study protein adsorption to flat solid interfaces, colloidal silica and latex beads, but its use to study protein adsorption at O/W interfaces is complicated by the significant amount of scattered light produced by small emulsion droplets (100 nm-10 µm). To overcome this problem and allow CD to be performed on emulsions Husband et al.15 developed a refractive index matched emulsion (RIME) method, in which the refractive index (RI) of the continuous phase is matched to that of the oil phase by adding glycerol and polyethylene glycol. In contrast to the FTIR measurements, RIME CD appeared to indicate an increase in α-helix content upon βLG adsorption to emulsion interfaces. These differences may be attributable to the aforementioned difficulty in obtaining CD spectra of proteins in emulsions. It is also possible that adding glycerol or polyethylene glycol may affect the protein conformation. Recently, however, synchrotron radiation CD confirmed the increase in α-helical content of β-LG upon adsorption to emulsion O/W interfaces18-20. Advanced instrumental methods allowing direct access to high resolution molecular information, especially changes in tertiary structure, are limited. Furthermore, for complex experimental systems, such as polydisperse emulsions, many different interfaces exist, with the protein responding to each local environment in a different way. One alternative, nonexperimental method that provides an opportunity to study the details of the protein conformational changes upon interaction with an O/W interface is molecular dynamics (MD) simulation. With the advancements of high performance computing, MD simulation serves as a
ACS Paragon Plus Environment
4
Page 5 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
powerful tool providing complementary information to experiments and allowing elucidation of the changes to protein secondary and tertiary structure stimulated by adsorption at an interface. Most simulation studies conducted to date have been confined to studies dealing with protein adsorption at liquid-solid interfaces21-24, protein adsorption on different self-assembled monolayers25, 26 or protein denaturation upon heating or addition of salts in solution27, 28. Recently, however, the interfacial and adsorption behaviour of proteins, surfactants and nanoparticles at different liquid-liquid interfaces has been studied using MD29-32. Both atomistic and coarse-grained simulations of barley lipid transfer protein adsorption at a decane-water interface showed the lipid transfer protein adsorbing with the helical region parallel to the interface. The average tilt angle normal to the interface was 73° for the all-atom model and 62° for the coarse-grained model31. The atomistic simulation showed the secondary structure of the protein to be conserved upon adsorption. MD simulations of nanoparticles (NPs) and nonionic surfactants at an O/W interface showed that at low surfactant concentration, NPs and surfactants have a synergistic effect in lowering the interfacial tension32. Adsorption of surfactant at the NP surface prevents NP aggregation, and also reduces the surfactant efficiency in lowering the surface tension between oil and water. A recent MD investigation of hydrophobin adsorption at an octane-water interface found adsorption to be essentially irreversible, and showed that surface structure and the flexibility of the protein have a large influence on the interfacial adsorption strength29. This study probed the adsorption of β-LG at an O/W (decane/water) interface using atomiclevel MD simulations, allowing the mechanism of adsorption to a purely hydrophobic surface and concomitant structural changes to the protein to be followed in detail. The ability of β-LG to
ACS Paragon Plus Environment
5
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 6 of 22
approach and interact with the decane surface, including its orientation with respect to the surface upon adsorption and the residues involved in the adsorption process were independent of its initial orientation with respect to the surface. The parts of β-LG that contribute to its final, stable adsorption on the decane surface were almost identical in all cases, however, driven by favourable van der Waals interactions between β-LG and decane. These interactions involve hydrophobic residues from the N-terminal region and the interior of the calyx, which become exposed due to significant rearrangement of the tertiary structure, although the secondary structure remains remarkably intact both during and after adsorption.
Methods Simulations All simulations were prepared and performed with the GROMACS simulation package version 4.5.433 and GROMOS 43A2 force field unless otherwise stated. Coulombic and van der Waals interactions were computed exactly within a cut-off distance of 1.0 nm. Outside of this, the longrange electrostatic interactions were treated using the particle mesh Ewald method34, 35 with a maximal spacing for the FFT grid of 0.16 nm and an interpolation order of 4. Coordinates and the topology file for a single decane molecule were obtained from S. Euston 31
. A slab comprising 818 decane molecules was built by random insertion of decane molecules
followed by energy minimisation, a 100 ps NVT simulation during which the temperature was increased from 50 K to 300 K, and finally a semi-isotropic 240 ps NpT simulation in which the area of the xy-surface was kept constant and the z-dimension allowed to vary until the experimental density of decane [720 g/L] was obtained. The final xyz dimensions of the oil slab were 8.56 × 8.56 × 3.43 nm, sufficient for the subsequent insertion and simulation of β-LG with
ACS Paragon Plus Environment
6
Page 7 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
cutoff distances of 1.0 nm and periodic boundary conditions. Prior to inserting β-LG, the dimensions in the z direction were increased to 12 nm so that the portion of the box into which the protein was inserted was the same size in all dimensions. Initial coordinates for β-LG, including crystal water molecules, were obtained from the X-ray structure (PDB ID: 3BLG). β-LG was added to the part of the system not occupied by the decane slab in three different orientations: rot0 (unmodified PDB coordinates, three replicates denoted rot0-1 etc); rot90x (rotated 90° around the x-axis, two replicates); rot90y (rotated 90° around the y-axis, two replicates) (Figure 1B). In all cases, the initial distance of the centre of mass of β-LG from the decane surface in the z-direction dinit was approximately 4 nm (Figure 1A). The simulation box was filled with SPC water molecules36 and the entire system subjected to energy minimization, heated from 50 K to 300 K over 100 ps in the NVT ensemble, and equilibrated for a further 1 ns in the NpT ensemble. The temperature and the pressure were maintained using the Berendsen thermostat and barostat37, with coupling times of τT = 0.1 ps and τp = 2 ps, respectively. The lengths of all bonds involving hydrogen atoms were constrained using LINCS38 with an order of 4, allowing for an integration time step of 2 fs. Each system was subsequently simulated for 200 ns, with coordinates saved every 5 ps. Analysis The majority of the analysis of the simulations was carried out using GROMACS analysis tools unless otherwise stated. The atom positional root mean square deviation (RMSD) from the initial crystallographic structure was calculated for the Cα atoms only after superimposing the Cα atoms onto their initial positions. The radius of gyration (Rg) was also calculated for the Cα atoms only. Secondary structure was calculated using the GROMACS do_dssp program, which
ACS Paragon Plus Environment
7
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 8 of 22
implements the algorithm of Kabsch and Sander39, and post-processed prior to plotting using a self-made Python script. The orientation angle θ was defined as the angle between the unit vector normal (n) to the decane surface (approximated by the z-axis) and the unit vector between the Cα atoms of residues 11 and 34, which approximates the long axis of β-LG in its natively folded structure (Figure 1A). The residues that come into contact with the oil slab were found using a self-made Tcl script in VMD40 to identify residues for which at least one atom was within dcontact = 0.6 nm of any atom of a decane molecule (Figure 1). The minimum distance between the protein and the decane slab was calculated using the GROMACS g_mindist program. The change in the hydrophobic surface area of β-LG was calculated using the GROMACS g_sas program without taking contact with decane into account, i.e. including the contribution from atoms that would be covered by the decane and so not solvent accessible.
Results and Discussion Approach of β-LG to the decane surface While in all cases, β-LG transports from solution (middle of the simulation box) to the oil surface, to which it eventually adsorbs, the time of approach varies both between replicates of a particular orientation of the protein with respect to the surface and between different orientations (Figure 2). For the three rot0 systems, β-LG approaches the decane surface very rapidly, within the first 20-30 ns, whereas the two rot90x replicates approach at times ranging from 60-100 ns (Figure 2A) Interestingly, rot90y-2 is among the first to approach the decane surface, whereas rot90y-1 is the last. Therefore the rate at which β-LG approaches the O/W interface is independent of its initial orientation, and is largely stochastic in nature.
ACS Paragon Plus Environment
8
Page 9 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
The initial contact between β-LG and the decane surface does not necessarily result in lasting contact formation and adsorption. Rot0-1 adheres to the surface quickly, with the number of protein residues in contact with the decane rapidly increasing to around 70 upon β-LG first contacting the surface, whereas rot90x-2 and rot90y-1, for instance, make multiple approaches to the surface before finally adsorbing at around 100 and 130 ns, respectively (Figure 2). Merely being close to the oil surface is therefore insufficient for adsorption to occur. To quantitatively characterize the approach of β-LG to the decane surface, the protein orientation, as defined in Figure 1A, was calculated throughout each simulation. During the first phase, when the β-LG is free in solution, the orientation angle θ undergoes significant fluctuation, ranging from ~60° in the case of rot90y-2 to almost 180° for rot90y-1 (Figure 2B), but the ultimate θ values once adsorption has occurred are in the range 90-135° in all cases. This is a surprisingly narrow range given the degree of deformation of the tertiary structure of β-LG upon adsorption (see Figure 4), indicating that the approach of the protein to the decane surface is not constrained by its initial orientation. In each simulation, the plateau in the value of θ coincides with β-LG contacting the decane surface (dcontact < 0.6 nm). The protein does not absorb immediately to the decane surface, but diffuses on the surface, as evidenced by the fluctuation in the x and y components and the overall magnitude |z| of the protein–decane centre of mass distance (Supporting Information Figures 1 and 2), but its orientation with respect to the surface does not change markedly during the adsorption process. Process of adsorption of β-LG to the decane surface
ACS Paragon Plus Environment
9
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 10 of 22
The convergence towards similar orientations of β-LG relative to the decane surface upon stable adsorption regardless of the initial orientation suggests that the same residues are involved in adsorption in all cases. To investigate whether successful adsorption requires particular residues of β-LG to come into contact with the oil surface, and where the adsorbing residues are located in protein structure, we identified the residues that are closest to the decane surface throughout the process of adsorption (Figure 3 and Supporting Information Figure 3). As discussed above, β-LG does not absorb instantly upon approaching the decane surface, but rather makes multiple approaches to the surface and diffuses laterally on the surface before finally adsorbing. During these initial phases, the number and identity of the residues in contact with the decane varies through time and between replicates (Figure 3 and Supporting Information Figure 3). Residues from the N-terminal region and around positions 50, 75, 100 (excluding the rot0 replicates), 110 (excluding rot90y-2) and 130-140 are generally the first to approach the decane. Within these regions there are specific residues, such as Lys77 and Lys14, which form contacts in almost all simulations. Later in the adsorption process, additional groups of residues around positions 16-35, 70-85 and 130-150 also come into contact with the decane, although again the exact residues involved differ between simulations. These later contacts are largely retained and thus can be considered to be drivers of stable adsorption. Notably, there are a large number of hydrophobic residues in the N-terminal region (Leu1, Ile2, Val3, Met7, Leu10, Ile12, Val15, Ala16, Leu22, Ala23, Met24, Ala25, Ala26, Ile29), and residues 70-123 correspond to the hydrophobic residues that line the calyx (β-strands E, F, D, G, Figure 3).
ACS Paragon Plus Environment
10
Page 11 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
Ultimately, ~80-90 residues are in contact with the decane surface once adsorption has taken place – approximately half of this 162-residue protein. Key regions that do not typically bind to the decane surface are the C-terminus (other than rot90y-2) and, in some but not all replicates, residues around 30-50, 60-70 and 120-130, with slight differences between the replicates in the exact location of the non-binding regions. These results indicate that while the initial contact between β-LG and the decane surface typically involves particular residues, e.g. Lys77 and Lys14, stable adsorption of β-LG to the decane surface is non-specific, driven by the predominantly hydrophobic N-terminal region and by rearrangement of the structure (Figure 5) to expose the hydrophobic residues that line the lipid-binding cavity. This is confirmed by the increase in the hydrophobic surface area (SASA) of the protein that occurs during the adsorption process (Figure 2E). The interactions between the newly exposed parts of the protein and the decane surface are energetically favourable, with the evolution of the protein-decane interaction energy closely tracking the approach of the protein to the surface (Figure 2). The strong correlation between the number of protein-decane contacts formed and the protein-decane interaction energy further indicates that the adsorption energy is not dominated by contributions from a few residues. Structural changes of β-LG upon adsorption The structural integrity of β-LG was monitored by calculating the atom-positional rmsd of the Cα atoms with respect to the initial structure of each simulation and the Rg (Figure 4A and B). The rmsd of a simulation of β-LG in solution without a decane slab present was also calculated for comparison. In all simulations with a decane slab present, the rmsd increases to be at least twice that of β-LG alone. The time at which the rmsd increases, however, differs between
ACS Paragon Plus Environment
11
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 12 of 22
simulations; in general, the rmsd increases somewhat later than the initial contact between β-LG and the decane surface occurs (Figure 2). The Rg values follow a similar trend to the rmsd, increasing some time after initial contact between β-LG and the decane surface. The nature of the changes in the overall shape of β-LG upon adsorption to the decane surface (Figure 5) is revealed in the decomposition of the Rg (Supporting Information Figure 4). The z component Rg,z in general tracks the overall Rg and increases throughout each simulation, whereas the x and y components are generally of lower magnitude. Somewhat counterintuitively, therefore, β-LG actually stretches in the z direction, i.e. normal to the decane surface, during the later stages of the adsorption process. Together, these results point to the structure of β-LG being affected little by the initial contact to the decane surface, and only undergoing significant change later, at the point of adsorption to the decane surface. The secondary structure content was also monitored (Figure 4C and D and Supporting Information Figure 5). While the tertiary structure of β-LG changes considerably upon its adsorption at the decane surface, the secondary structure is well conserved throughout these structural changes, with close to 100% of the protein retaining its initial classification into a given type of regular structure (Figure 4C and D, Figure 5 and Supporting Information Figure 5). Adsorption to the surface therefore results in rearrangement of secondary structure elements, rather than a complete loss of structure. This is in contrast to the results of Fang and Dalgleish14, who observed a decrease in intra-molecular β-sheet content upon adsorption. This was accompanied by a concomitant increase in inter-molecular β-sheet content, something that cannot be investigated in single molecule simulations. The increase in α-helix observed with
ACS Paragon Plus Environment
12
Page 13 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
RIME CD15 and synchrotron radiation19, 20 is also not observed here, most likely due to the limited time scale of the simulations compared to the experimental time scale. While coarsegrained simulations would allow longer time scales to be accessed, they are inappropriate for probing structural changes due to the common requirement for secondary structure constraints. It will be of interest, therefore, to see whether additional experimental studies can resolve the current discrepancy between the secondary structure changes observed by RIME CD and FTIR. Matched computational and experimental studies, such as using steered MD simulations to mimic the single-molecule force spectroscopy that has been carried out for β-LG at an O/W interface42, also hold much promise. Conclusions The adsorption of proteins to O/W interfaces occurs in many biological processes and is critical to the physicochemical properties of functional foods, particularly those based on emulsions stabilised by milk proteins. While low-resolution structural properties of adsorbed proteins can be measured experimentally, detailed characterisation of the process of adsorption and concomitant structural changes requires an alternative approach. Here, MD simulations of βLG in the presence of a decane slab allowed the approach and adsorption of β-LG to the O/W interface to be followed in atomic-level detail. The approach phase of the adsorption was largely stochastic in nature. The residues that make contact with the decane surface and the final orientation of β-LG with respect to the surface are independent of the initial orientation of β-LG. The adsorption process was found to be largely driven by groups of predominantly hydrophobic residues, some of which become exposed during adsorption by structural rearrangements. These structural changes preserve the secondary structure of β-LG, which is in contrast to some experimental results; these discrepancies are most likely due to differences between the nature
ACS Paragon Plus Environment
13
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 14 of 22
and time-scale of the experimental and computational studies. The detailed characterisation of the adsorption of β-LG at an O/W interface presented here, particularly the identification of the residues that drive the adsorption process and their location in the native structure of β-LG will enable improved engineering of functional food products and drug delivery systems.
Supporting Information. β-LG-decane distance, β-LG adsorption landscapes, distance to decane of each residue during adsorption, Rg of β-LG, secondary structure content of β-LG. This material is available free of charge via the Internet at http://pubs.acs.org.
ACS Paragon Plus Environment
14
Page 15 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
Figure 1. Schematic diagrams of the simulation system. A) Definition of the protein-surface orientation angle θ and the protein-surface separation distances dinit and dcontact. B) An example of the rotation procedure used to generate the different initial orientations.
ACS Paragon Plus Environment
15
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 16 of 22
Figure 2. Approach of β-LG to the decane surface. Time-series of A) minimum distance between β-LG and decane surface; B) orientation angle θ between β-LG and the decane surface; C) number of β-LG residues in contact with decane surface; D) interaction energy between β-LG and decane and E) hydrophobic surface area. The protein-decane potential energy is used as a
ACS Paragon Plus Environment
16
Page 17 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
proxy for the adsorption energy; only the van der Waals energy contributes, as the electrostatic component is zero due to the decane molecules having zero partial charge.
Figure 3. Residues involved in β-LG adsorption. Contact between a β-LG residue and the decane surface (dcontact < 0.6 nm) is indicated in blue (hydrophilic) or red (hydrophobic). The eight β-strands and two α-helices of β-LG are labelled on the structure shown in the top right panel and their location in the sequence indicated by the vertical panels.
ACS Paragon Plus Environment
17
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 18 of 22
Figure 4. Structural properties of β-LG. Time-series of the A) atom-positional rmsd of the Cα atoms with respect to the initial structure; B) Rg; C) percentage of residues exhibiting each type of secondary structure; and D) secondary structure of each residue during the simulations. Panels C) and D) show data for the rot0-1 simulation only; the secondary structure content of β-LG in the other simulations is similarly constant throughout each simulation (Supporting Information Figure 5).
ACS Paragon Plus Environment
18
Page 19 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
Figure 5. Structural changes to β-LG upon adsorption. Snapshots of β-LG at the start of each simulation (top) as labelled, and at the end, after adsorption to the decane surface (bottom).
ACS Paragon Plus Environment
19
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 20 of 22
Corresponding Author * Correspondence should be addressed to Jane R. Allison, [email protected]. Author Contributions The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Funding Sources JRA received financial support from a Massey University Research Fund Early Career Grant and a Marsden Fund Fast Start award (13-MAU-039).
The authors wish to acknowledge Bluefern (University of Canterbury) for provision of supercomputing facilities, and Stephen R. Euston (Herriot-Watt University) for providing coordinate and topology files for decane.
β-LG, β-lactoglobulin; O/W, oil/water; Rg, radius of gyration; rmsd, root-mean-square deviation.
ACS Paragon Plus Environment
20
Page 21 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
References (1) Zhai, J. l.; Day, L.; Aguilar, M.-I.; Wooster, T. J. Curr. Opin. Colloid Interface Sci. 2013, 18, 257. (2) Singh, H. Food Hydrocolloids 2011, 25, 1938. (3) Thompson, A.; Boland, M.; Singh, H.; (eds) Milk Proteins: from Expression to Food; 2nd ed.; Academic Press, Elsevier Inc., 2009. (4) Brownlow, S.; Cabral, J. H. M.; Cooper, R.; Flower, D. R.; Yewdall, S. J.; Polikarpov, I.; North, A. C. T.; Sawyer, L. Structure 1997, 5, 481. (5) Kontopidis, G.; Holt, C.; Sawyer, L. J. Mol. Biol. 2002, 318, 1043. (6) Qin, B. Y.; Bewley, M. C.; Creamer, L. K.; Baker, H. M.; Baker, E. N.; Jameson, G. B. Biochemistry 1998, 37, 14014. (7) Uhrínová, S.; Smith, M. H.; Jameson, G. B.; Uhrín, D.; Sawyer, L.; Barlow, P. N. Biochemistry 2000, 39, 3565. (8) Kumosinski, T. F.; Timasheff, S. N. J. Am. Chem. Soc. 1966, 88, 5635. (9) Mckenzie, H. A., Sawyer, W. H Nature 1967, 214, 1101. (10) Sakurai, K.; Oobatake, M.; Goto, Y. Prot. Sci. 2001, 10, 2325. (11) Dalgleish, D. G. Food Hydrocolloids 2006, 20, 415. (12) Kim, H. J.; Decker, E. A.; McClements, D. J. J Agric Food Chem 2002, 50, 7131. (13) Kim, H. J.; Decker, E. A.; McClements, D. J. Langmuir 2002, 18, 7577. (14) Fang, Y.; Dalgleish, D. G. J. Colloid Interface Sci. 1997, 196, 292. (15) Husband, F. A.; Garrood, M. J.; Mackie, A. R.; Burnett, G. R.; Wilde, P. J. J Agric Food Chem 2001, 49, 859. (16) Lee, S.-H.; Lefèvre, T.; Subirade, M.; Paquin, P. J Agric Food Chem 2007, 55, 10924. (17) Sakuno, M. M.; Matsumoto, S.; Kawai, S.; Taihei, K.; Matsumura, Y. Langmuir 2008, 24, 11483. (18) Day, L.; Zhai, J.; Xu, M.; Jones, N. C.; Hoffmann, S. V.; Wooster, T. J. Food Hydrocolloids 2014, 34, 78. (19) Zhai, J.; Miles, A. J.; Pattenden, L. K.; Lee, T.-H.; Augustin, M. A.; Wallace, B. A.; Aguilar, M.I.; Wooster, T. J. Biomacromolecules 2010, 11, 2136. (20) Zhai, J.; Wooster, T. J. Langmuir 2011, 27, 9227. (21) Zhao, D.; Peng, C.; Zhou, J. Phys. Chem. Chem. Phys. 2015, 17, 840. (22) Hagiwara, T.; Sakiyama, T.; Watanabe, H. Langmuir 2009, 25, 226. (23) Wei, T.; Carignano, M. A.; Szleifer, I. Langmuir 2011, 27, 12074. (24) Wei, T.; Carignano, M. A.; Szleifer, I. J. Phys. Chem. B 2012, 116, 10189. (25) Liu, J.; Liao, C.; Zhou, J. Langmuir 2013, 29, 11366. (26) Peng, C.; Liu, J.; Zhao, D.; Zhou, J. Langmuir 2014, 30, 11401. (27) Euston, S. R. Food Hydrocolloids 2013, 30, 519. (28) Euston, S. R.; Ur-Rehman, S.; Costello, G. Food Hydrocolloids 2007, 21, 1081. (29) Cheung, D. L. Langmuir 2012, 28, 8730. (30) Cheung, D. L. Adv. Chem. 2014, 2014, 12. (31) Euston, S. R. Biomacromolecules 2010, 11, 2781. (32) Ranatunga, R. J. K. U.; Nguyen, C. T.; Wilson, B. A.; Shinoda, W.; Nielsen, S. O. Soft Matter 2011, 7, 6942. (33) Hess, B.; Kutzner, C.; van der Spoel, D.; Lindahl, E. J. Chem. Theory Comput. 2008, 4, 435. (34) Darden, T.; York, D.; Pedersen, L. J. Chem. Phys. 1993, 98, 10089. (35) Essmann, U.; Perera, L.; Berkowitz, M. L.; Darden, T.; Lee, H.; Pedersen, L. G. J. Chem. Phys. 1995, 103, 8577. (36) Berweger, C. D.; van Gunsteren, W. F.; Müller-Plathe, F. Chem. Phys. Lett. 1995, 232, 429. (37) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; DiNola, A.; Haak, J. R. J. Chem. Phys. 1984, 81, 3684.
ACS Paragon Plus Environment
21
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 22 of 22
(38) Hess, B.; Bekker, H.; Berendsen, H. J. C.; Fraaije, J. G. E. M. J. Comput. Chem. 1997, 18, 1463. (39) Kabsch, W.; Sander, C. Biopolymers 1983, 22, 2577. (40) Humphrey, W.; Dalke, A.; Schulten, K. J. Mol. Graph. 1996, 14, 33. (41) Pieranski, P. Physical Review Letters 1980, 45, 569. (42) Touhami, A.; Alexander, M.; Kurylowicz, M.; Gram, C.; Corredig, M.; Dutcher, J. R. Soft Matter 2011, 7, 10274.
TOC Graphic
ACS Paragon Plus Environment
22
