Insights from Molecular Dynamics Simulations of Albumin Adsorption on Hydrophilic and Hydrophobic Surfaces

Abstract

Protein adsorption at the surface affects the material biocompatibility directly as it is the first reaction that happens when a foreign material comes in contact with blood. In this study, the mechanism of albumin adsorption on hydrophilic and hydrophobic surfaces is investigated. Although it is studied extensively and has been of keen interest for decades, the adsorptive nature of albumin is still not fully understood with contradicting reported studies. This problem results from previous works focusing on mostly qualitative and quantitative adsorption properties of albumin, rather than the specific interaction mechanisms. The variable local surface properties across albumin can significantly impact adsorption and must be explored. In this work, the effect of hydration is found to significantly increase adsorption with minor reductions. The adsorption of albumin on hydrophilic or hydrophobic surfaces is dependent on albumin orientation, which is dictated by local charge effects. Based on these findings, an optimized material surface is proposed to minimize albumin adsorption using functional groups to limit surface availability for hydrophobic interactions while inhibiting excess electrostatic effects at hydrophilic sites. The extent of albumin adsorption and shape change are characterized herein using the heat capacity. Current study identifies interaction mechanisms previously missing in literature, which are responsible for inconsistent adsorption results.

Graphical Abstract

1. Introduction

Hemocompatibility, a key factor in the design of implantable and other biomedical devices, is a thrombotic response to a surface in contact with blood [1]. It pertains to a wide range of materials suitable for bioapplication such as needles for vascular graft and heart valves [2], load bearing for spinal disks [3], membranes for blood filtration [4], etc. Generally, the mechanism of host reaction to a biomaterial is very complicated and involves many different processes such as protein adsorption and deformation, platelets adhesion and complement activation, and provisional matrix formation. When a biomaterial is introduced to blood, depending on the protein surface properties, instantaneous protein biofouling may occur and trigger an inherent biological response [1] to a foreign body invasion, which include a series of reactions such as coagulation and platelet adhesion. Surface modification and blending specific compounds with the base material are known methods to improve hemocompatibility and several efforts have been made to utilize them for improving biocompatibility of materials [4–11]. Plasma protein adsorption is the first and fastest reaction that occurs and results in a set of subsequent responses [12–14]. Human serum albumin (HSA) is an important protein in blood with various important functionalities [15–18] and forms 60% of whole blood proteins. Therefore, understanding of its interaction with surfaces provides good insight into the material’s hemocompatibility. Generally, fibrinogen and other proteins mediate the platelet adhesion to surfaces while HSA is considered inert [14] because its adhesion to a surface could prevent platelet adhesion [19].

Hydrophilicity and hydrophobicity of materials are the key factors known to impact protein-protein interaction [20] and protein adsorption on the surface which can be tuned by surface modification [21–24]. Many studies suggest that hydrophilic surfaces and compounds in the base material through blending [25] or addition [26–29] reduce HSA adsorption while hydrophobic surfaces increase it. Jesus et al. [30] studied the adsorption of HSA on surfaces of hydrophobic and hydrophilic Ti6A14V powder and reported lower protein adsorption on hydrophilic surfaces. Oss et al. [31] investigated the interaction between proteins and inorganic oxides and mentioned that silica has a low HSA adsorption due to its hydrophilicity and a negatively charged surface measured by microelectrophoresis [32] that repels HSA. Mücksch and Urbassek [33] studied the effect of surface hydrophobicity on bovine serum albumin (BSA), which has similar surface properties to HSA [34], and indicated that BSA adsorbs to the surface, partially unfolds, and experiences denaturation.

On the contrary, some studies suggest that hydrophobic surfaces have lower amounts of adsorption. Rupert et al. [35] studied the BSA adsorption on different surfaces and showed that at the low BSA concentration, 0.1 mg mL⁻¹, the most hydrophobic surface has the lowest amount of BSA adsorption on the surface. However, at a high BSA concentration of 10 mg mL⁻¹, which is closer to that of HSA in blood i.e. 35–50 mg mL⁻¹, a slightly hydrophilic surface results in minimum adsorption. Supported by simulation, Jeyachandran et al. [36] showed experimentally that the interaction forces between BSA and hydrophilic surfaces are stronger than that of the hydrophobic one and ~95% of the hydrophilic surface is covered by BSA, whereas it is ~50% for the hydrophobic one. Absolom et al. [37] reported that siliconized glass made by treating glass surface with silicon oil according to the Neumann and Renzow method [38] has slightly higher hydrophobicity than Teflon and results in lower protein adsorption. It is hypothesized that it is attributed to the “screening” phenomenon where the adsorbed protein on surface changes conformationally, effectively screening new proteins. Sousa et al. [39] reported the same behavior for hydrophobic TiO₂ sp versus hydrophilic TiO₂ cp where it shows a lower work of adhesion and easier exchange of adsorbed HSA with solution. Finally, some studies mention that the adsorption happens on both surfaces regardless of the surface type, such as a study by Malgoratza et al. [40].

Despite the extensive studies on the effect of surface hydrophilicity and hydrophobicity on HSA adsorption, the reported results are inconsistent and the underlying mechanism of this process is not clearly known. It is pertinent to mention that HSA is a large protein with various folded hydrophobic sites inside and charged hydrophilic sites on surface [41,42] that interact with a surface, resulting in different adsorption properties depending on the direction in which it approaches the surface. However, the focus of prior studies have been on adsorption properties of HSA regardless of its orientation and local surface properties. In this study, the local surface properties of HSA at the molecular level are considered and it is shown how hydration and surface charge affect its adsorption. Moreover, the effect of different orientations of HSA when approaching various surfaces are evaluated. In each case, the binding energy between the protein and surface is calculated and the heat capacity variation [43] is used to evaluate its shape change and extent of adsorption. This work provides a better understanding of blood protein interaction with various surfaces such as GO biocompatible membranes, which has precise molecular cut-off and 40 times higher permeability relative to high flux dialysis membranes [44].

2. Setup and Methodology

2.1. Simulation box

Simulations are performed by Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS) [45]. In each simulation there is one HSA molecule, water molecules, and a graphene or graphene oxide sheet. In each case, a set of simulations are performed in six directions for albumin to evaluate its interaction with different surfaces from all facets it approaches them. These directions are the original position in the data file, +90X, −90X, +180X, −90Y, and +90Y where the sign shows the rotation direction using right hand rule and the letter shows the axis of rotation. Graphene and graphene oxide (GO) sheets are used as hydrophobic and hydrophilic surfaces, respectively, and positioned in the bottom of the box i.e. −z direction. Three different levels of graphene oxide is used to investigate the effect of oxidation on the results. For all simulations, periodic boundaries in x and y-directions, fix boundary in −z, and fix-wall command in the +z-direction are applied. The fixed boundary and fix-wall command in the +z direction allows HSA to interact with the given surface only inside the simulation box and not through the boundary. Length (L_x), width (L_y), and height (L_z) of the simulation box are 10 nm, which is large enough to avoid any self-interaction of HSA through periodic boundaries. Figure 1 shows a simulation box with HSA surrounded by water molecules above a graphene sheet.

Figure 1.

Human serum albumin molecule in original direction (base case), water molecules, and graphene as a hydrophobic surface in simulations (as seen in VMD software).

2.2. Ensembles and Simulation Parameters

Graphene and GO sheets are fixed and NVT ensemble is used for 0.1 ns to equilibrate water molecules and HSA. NPT ensemble is used for 10 ns to run simulations with a time step of 1 fs. Unless otherwise stated, Nose-Hoover thermostat is used to control the temperature at 310 K, which is equal to the temperature of blood or 37 °C. Real units, periodic boundaries, bond style of harmonic, angle and dihedral styles of hybrid consisting of charm and harmonic, and improper style of harmonic are used. Cut-off of 12 Å is applied for both Lenard Jones and columbic terms. To model the non-bonded interatomic interactions, the Lenard Jones (LJ) model of 6–12 is used as follows [46]:

$${\phi }_{ij}=4{\epsilon }_{ij}\left\{{\left(\frac{{\sigma }_{ij}}{r_{ij}}\right)}^{12}-{\left(\frac{{\sigma }_{ij}}{r_{ij}}\right)}^6\right\}$$

where φ_ij is the LJ potential, ε_ij is the potential well depth, σ_ij is the equilibrium distance between atoms, and r_ij is distance between atoms.

2.3. Data Files and Models

The SPC/E water model explained by Berendsen et al. [47] with an initial density of 1 g/cm³ is used in the simulations. The graphene and GO data files were made according to the existing structures and models [48–57]. The graphene sheet is a 2D hexagonal lattice of carbon atoms with a bond length of 1.4 Å. Graphene and GO layers and functional groups are fixed in all simulations and no interaction is considered between the atoms in a sheet. The hydrophilic groups on GO sheets are on the edges and defects where carbon has free electrons [58]. Thus, to make the GO sheets from graphene, an epoxy or hydroxyl group is introduced to each carbon atom at defects in the form of a dual or single bond, respectively. Three levels of oxidations are used in simulation by removing every 3, 6, and 9 carbon atoms to make a defect.

The water model and a schematic of G and GO structures with the functional groups on GO are shown in Figure 2. To generate different levels of graphene oxide one carbon atom is removed every three, six, and nine atoms, which corresponds to high, normal, and low oxidation levels respectively, and is replaced by functional groups. Among various interatomic potentials between carbon and water molecules, the one with simulation results that best matched the experimental properties, such as contact angle, was selected for Carbon-Water interactions [59].

Figure 2.

Water model (SPC/E), graphene, and functional groups.

In this model, σ_C-O = 3.19 Å and ε_C-O = 0.0937 kcal/mol. The parameters for water SPC/E [47] model are used to account for Water-Water interactions. The interaction parameters between atoms within the functional groups are adopted from a study by Jiao et al. [60]. The bond length and angle coefficients are adopted from Levitt et al. [61]. These parameters are summarized in Table 1. For those interactions that are not explicitly set, the mix arithmetic pair modify function is used for LJ parameters as ${\sigma }_{ij}=\frac{\left({\sigma }_i+{\sigma }_j\right)}{2}$ and ${\epsilon }_{ij}=\sqrt{{\epsilon }_i\times {\epsilon }_j}$ [62]. This method was successfully used previously [63]. HSA data and topology files were downloaded from Protein Data Bank (RCSB PDB) [64] and used in VMD [65] to generate a LAMMPS data file. Charmm36 force-field [66,67] along with cmap36 file necessary for proteins [68,69] are adopted.

Table 1.

Interaction parameters. σ_C-O and ε_C-O are defined separately.

Molecule	Atom	q (e)	σ (Å)	ε (kcal/mol)	b0 (Å)	kb (kcal/mol)	θ (°)	Kθ (kcal/mol)
Graphene	C	0.0	3.83	0.0551	C-C, 1.4	C-C, 250	C-C-C, 120	C-C-C, 50
Epoxy	C	+0.2	3.4	0.0693	C-O, 1.41	C-O, 450	C-C-O, 90	C-C-O, 50
	O	−0.4	2.9	0.142
Hydroxyl	C	+0.1966	3.4	0.0693	C-O, 1.41	C-O, 450	C-C-O, 90	C-C-O, 50
	O	−0.5260	3.166	0.14	O-H, 0.945	O-H, 450	C-O-H, 110	C-O-H, 50
	H	+0.3294	0.0	0.0
Water	O	−0.8476	3.166	0.1553	O-H, 1.0	O-H, 450	H-O-H, 109.47	H-O-H, 55
	H	+0.4238	0.0	0.0

HSA has a globular protein composed of 585 amino acids with molecular weight of 68000 Da [69]. About 7% of the blood is protein and albumin forms almost 2/3 (about 60%) of the whole blood protein volume [18,70]. HSA has a heart-like shape and is about 80 Å on the side with an average thickness of about 30 Å. Its volume is 88249 ų in non-solvated state and increases up to 20% when solvated in water [17].

2.4. Methodology

The binding energy and heat capacity are used to evaluate interactions of the HSA with the surface and its shape change, respectively, as described below.

1. Binding Energy: To calculate the binding energy, two atom groups of “HSA” for albumin and “Surface” for graphene or graphene oxide are defined. Using group/group command in LAMMPS, the total energy and force interaction between these two groups are calculated.

2. Heat Capacity: To calculate the heat capacity, two methods are utilized:

   • Total energy change with temperature: $C_p={\left(\frac{\partial q}{\partial T}\right)}_p={\left(\frac{\partial E}{\partial T}\right)}_p$
     where q and E are heat and total energy of the system, respectively.
   • Fluctuation of potential energy [71]: $C_p=\frac{\langle E^2\rangle -{\langle E\rangle }^2}{k_b{T}^2}$
     where 〈 〉, E, and k_b are ensemble average, total energy, and Boltzmann constant, respectively.

3. RESULTS AND DISCUSSION

First, the effect of HSA hydration on its adsorption to the surface was investigated and it was shown that reducing the hydration level increases the adsorption significantly. Then, the effect of surface hydrophobicity and hydrophilicity with different oxidation levels on adsorption of HSA is examined and the binding energy for each case is calculated. It is shown that the binding energy correlates with surface charge directly. Finally, the heat capacity methods, validated by calculating water heat capacity, are used to investigate the shape change of HSA.

3.1. Effect of Hydration on Albumin Adsorption

HSA adsorption on a hydrophobic surface requires dehydration between HSA and surface. Simulation of HSA on graphene as a hydrophobic surface reveals a weak binding between them, which is expected as the hydrophobic sites are inside HSA and hydrophilic sites are covered with water molecules on the surface. Results show that any reduction in the number of water molecules around HSA result in higher adsorption on the surface. Partial dehydration of HSA can be achieved by either reducing water content in the simulation box or adding ions or any other molecules to water as they result in dehydration by decreasing the number of water molecules available for HSA. This explains why higher concentrations increase adsorption on hydrophobic surfaces [34]. Results show that reducing the HSA hydration by ~10% result in ~2.5 times stronger adsorption on the surface, while the effect on water adsorption and its binding energy at the surface were negligible. Table 2 shows the ratio of binding energy among water, surface, and albumin for partial dehydration over full hydration. In full hydration the empty space inside the simulation box is full of water molecules with bulk density of 1g/cm³, while in reduced conditions the water content is decreased by 20%. The provided data are the averaged values for six different facets of HSA approaching the hydrophobic surface of graphene. The provided error bars show ~21% error for binding energy between HSA and graphene using three different results for each facet, while it is negligible for the other two.

Table 2.

Binding energy ratio of HSA on surface of graphene (partial dehydration over full hydration) and its standard error.

Binding Surfaces	Binding Energy Ratio, Cal/mol	Standard Error
HSA-Water	0.894	0.006
HSA-Graphene	2.501	0.213
Graphene-Water	0.997	0.002

The same simulations for graphene oxide reveal that the hydrophilic-hydrophilic interaction between different sites on the surface of HSA and functional groups on graphene oxide is strong enough to dehydrate the surface easily. The rest of simulations in this study are performed with reduced water content to allow for HSA adsorption as it occurs in real conditions in the presence of ions and other species in blood. The lower water numbers decrease the computational costs as well.

3.2. HSA adsorption on hydrophilic/hydrophobic surface

Binding energy between the surface and HSA molecules is used as a scale to compare the interaction forces between them in different conditions. Results show that HSA adsorbs on all hydrophobic and hydrophilic surfaces regardless of the local surface charge and oxidation type, however the binding energy magnitude is a function of the above-mentioned parameters. Figure 3 shows the results of binding energy for HSA on the hydrophilic surface of graphene oxide with an average oxidation level in all six directions.

Figure 3.

Binding energy between HSA and hydrophilic graphene oxide with average oxidation level. Different facets of HSA results in different binding energy values.

The same simulations are performed for all surfaces in six directions and repeated three times. The average value for all cases are summarized in Table 3.

Table 3.

Average values of binding energy for HSA and different surfaces

HAS Direction\Type	G	GOLow	GOAve	GOHigh
+90X, kcal/mol	−32	−4	−27	−115
+90Y, kcal/mol	−108	−229	−73	−211
+180X, kcal/mol	−177	−114	−99	−445
Base Case, kcal/mol	−84	−20	53	28
−90X, kcal/mol	−67	−109	−121	−319
−90Y, kcal/mol	−96	−72	−150	−78
Average, kcal/mol	−94	−91	−69	−190

According to Table 3, the binding energy starts with a moderate value at the hydrophobic surface of graphene, reduces over the low oxidized graphene, and becomes minimum over an average oxidation level. This behavior is reported by Rupert et al. where they have examined the BSA adsorption on different surfaces and have mentioned that at high concentrations, close that of HSA in blood, the adsorption on slightly hydrophilic surfaces is minimum [35]. However, the adsorption increases significantly at high oxidation level, which can be compared to a study by Jeyachandran et al. where they have reported that 95% of their hydrophilic surface was covered by BSA [36]. The strength and nature of interaction is a function of both surface and protein properties. When the surface is hydrophobic, the dehydration of surface and hydrophobic-hydrophobic interaction is the determining factor and binding energy gradually increases over time as dehydration progresses. However, in the presence of surface charges, i.e. hydrophilic surfaces of graphene oxide, electrostatic forces are dominant and the binding energy increases with oxidation rate. Figure 4 shows the net charge of the HSA surface at two extreme situations in terms of adsorption which are base case and the one rotated 180 degree around X i.e. +180X as a function of involved length of HSA starting from its closest atom to GO surface. The positive surface charge of +180X and negative charge of the base case are good indicators of the charge dominance on adsorption processes as the surface of graphene oxide has many oxygen with negatively charged electron clouds that adsorb the positive charges and repel the negative ones.

Figure 4.

Net charge of HSA surface as a function of its involved length starting from its closest atom to the hydrophilic surface of graphene oxide.

The results of binding energy indicate that a surface with a moderate level of oxidation minimizes the HSA adsorption, while both hydrophobic and highly hydrophilic surfaces increase the adsorption. The reason behind this is that the slightly oxidized graphene results in a weaker hydrophilic-hydrophilic interaction compared to highly oxidized one and at the same time it limits the availability for GO hydrophobic domains and avoids hydrophobic-hydrophobic interaction.

3.3. HSA shape change on hydrophilic/hydrophobic surface

During HSA adsorption, protein shape can change and result in its denaturation which may lead to malfunctioning of the protein. In order to evaluate shape change, two methods of heat capacity were used and they have been validated by calculating heat capacity of water prior to apply them to HSA. Figure 5 shows the results of water heat capacity simulations for both methods, which are in good agreement with experimental values.

Figure 5.

Validation of heat capacity method by calculating heat capacity of water. Both methods are in good agreement with experimental water heat capacity of 1 cal/g.K.

These methods are used for solvated HSA in water and they both show close results as shown in Figure 6. While HSA heat capacity is below that of bulk water at 1 J/mol.K, results show that it increases significantly by about 8 times in the presence of both hydrophobic graphene and hydrophilic graphene oxide surfaces. One reason behind this is that the added heat to the HSA will be adsorbed to overcome the interaction between HSA hydrophilic sites and both water and hydrophilic molecules on the surface of graphene oxide leave no energy for the albumin shape change as seen in simulations.

Figure 6.

Heat capacity of HSA in water using total energy method (on the left) and fluctuation of energy around average energy in ensemble (on the right). Results show 10% difference.

Another reason for this behavior is the protein denaturation process under different conditions, such as protein shape changes, which is dominant on the hydrophobic surface of graphene where hydrophobic-hydrophobic interactions unfold the protein. Heat capacity increase during unfolding is reported in different studies previously. Privalov et al. have shown how a protein’s heat capacity increases with temperature [43], Cooper has reported a significant increase in heat capacity of a small globular protein (lysozyme) during unfolding [72], and Myers et al. investigated 45 proteins and reported the increase in heat capacity of these proteins upon unfolding. This sudden increase in heat capacity is the same as the melting process in other materials such as water where the added heat is used to perform the phase change. The calculated heat capacity value for all cases of graphene and graphene oxide with different oxidation levels are almost the same with a maximum of 20% variation. Figure 7 shows the result of heat capacity of HSA on graphene and graphene oxide with an average oxidation level.

Figure 7.

Heat Capacity of HSA on the surface of graphene on the left and graphene oxide with average oxidation level on the right side.

4. CONCLUSION

This study sheds light on the underlying mechanism of human serum albumin adsorption and shape changes on hydrophilic and hydrophobic surfaces using molecular dynamics simulations. To evaluate the adsorption of HSA on a surface the binding energy method was used to show that the adsorption is a function of hydration degree and local surface properties of the HSA facet, such as charge, when it approaches the material surface. The protein surface properties were used to introduce an optimum level of oxidation that minimizes the average adsorption of HSA by limiting the hydrophobic surface availability and minimizing hydrophilic interaction. In addition, the shape change was investigated using heat capacity variation methods validated by calculating water heat capacity. It was shown that the heat capacity of hydrated HSA is lower than that of water, but it increases significantly in the presence of any surfaces which is an indication of high binding energy between the protein and hydrophilic surfaces and its denaturation on hydrophobic surfaces. The current study provides a good insight on the mechanism of HSA adsorption on different surfaces and reveals how surface properties of protein affect the adsorption process and how these properties can be utilized to optimize the adsorption and control the protein deformation.

• Slight albumin dehydration increases the interaction energy with surface significantly.

• Hydrophilic surface increases binding energy, but doesn’t deform albumin significantly.

• Hydrophobic surface deforms the protein and results in protein denaturation.

• Binding energy is minimum on a moderately hydrophilic surface.

• Net charges of the surface and albumin are dominant factors in adsorption behavior.

DATA AVAILABILITY

The data support the findings on this study are available from the corresponding author upon reasonable request.