Rickard
Frost
*ab,
Christoph
Langhammer
b and
Tommy
Cedervall
c
aDepartment of Energy and Environment, Chalmers University of Technology, SE-412 96, Gothenburg, Sweden. E-mail: rickard.frost@chalmers.se
bDepartment of Physics, Chalmers University of Technology, SE-412 96, Gothenburg, Sweden
cBiochemistry and Structural Biology and NanoLund, Lund University, Box 124, SE-221 00 Lund, Sweden
First published on 23rd February 2017
Biomolecules such as proteins immediately adsorb on the surface of nanoparticles upon their exposure to a biological environment. The formed adlayer is commonly referred to as biomolecule corona (biocorona) and defines the biological activity and toxicity of the nanoparticle. Therefore, it is essential to understand in detail the biocorona formation process, and how it is governed by parameters like composition of the biological environment, and nanoparticle size, shape and faceting. Here we present a detailed equilibrium and real time in situ study of biocorona formation at SiO2-nanoparticle surfaces upon exposure to defined (BSA, IgG) and complex (bovine serum, IgG depleted bovine serum) biological samples. We use both nanofabricated surface-associated Au core–SiO2 shell nanoparticles (faceted, d = 92–167 nm) with integrated nanoplasmonic sensing function and dispersed SiO2 nanoparticles (using DLS and SDS-PAGE). The results show that preadsorbed BSA or IgG are exchanged for other proteins when exposed to bovine serum. In addition, the results show that IgG forms a biocorona with different properties at curved (edge) and flat (facet) SiO2-nanoparticle surfaces. Our study paves the way for further real time in situ investigations of the biocorona formation and evolution kinetics, as well as the role of molecular orientation in biocorona formation, on nanoparticles with surface faceting.
The formation of the biocorona (hereafter referred to as corona) is typically a dynamic process in which different biomolecules compete for the available surface.9–11 Proteins with high concentration will likely bind first but with time be replaced by proteins with high affinity for the surface.12 The on/off rates and the concentration of each biomolecule will determine the contents in the corona over time until equilibrium is reached. For example, there are a number of studies describing the multiple protein binding to silica nanoparticles in complex protein mixtures,5,10,11,13–15 whereas another recent study reports that in human blood plasma a single protein is dominating when the particle surface area is scarce.11 An alternative view of the formation of the corona is that biomolecules stochastically adsorb irreversibly to the nanoparticle surface.16 It is likely that the nature of the nanoparticle surface determines which process dominates and that on some surfaces a combination of different processes occurs. However, very little is known about correlations between simultaneously occurring molecular adsorption processes and specific nanoparticle descriptors such as size and shape because real time corona formation studies are experimentally very difficult.
Commonly, when the corona formed in blood serum or plasma is studied, nanomaterials (often in the form of particles) are incubated in biological media. The formed complexes of the nanomaterial with adsorbed proteins, lipids or other biomolecules, are separated from unbound biomolecules (e.g. by centrifugation), and characterized (e.g. by mass spectrometry or gel electrophoresis). In this way, it is possible to identify the constituents of the corona. However, the described methodology has three main drawbacks: (1) the nanomaterials may aggregate upon incubation in biological media, (2) the formed corona may change during the purification process, and (3) only the equilibrium state is probed. Thus, the result of the final protein characterization may be affected in the sense that it not fully represents the initially formed corona. Another approach is to measure the size of the biomolecule/nanomaterial complex over time by dynamic light scattering (DLS), as the size of monodisperse complexes is related to the thickness of the corona. DLS analysis may be performed during incubation in biological media, thus purification of the nanoparticles may not be necessary if it is possible to distinguish between the response from the nanoparticles and the biological medium. However, it can be difficult to separate aggregation from the increase in hydrodynamic radius, which means that the result often is largely biased towards bigger entities. An alternative method is nanoparticle tracking analysis (NTA). However, in NTA there is often a need to dilute the samples before measurement, which again may change the equilibrium composition of the corona. The third method often used to measure the nanoparticle size, and to follow changes thereof is differential sedimentation centrifugation (DSC). The method is robust, but in mixed samples (of e.g. biomolecules and nanoparticles), the observed size of protein/nanoparticle complexes is difficult to assess, as the density of the protein/nanoparticle complexes is unknown.
As a complementary technology, to address some of the aforementioned shortcomings of existing techniques, we recently introduced a nanoplasmonic sensor (NPS) surface.17 This technique enables real-time in situ analysis of corona formation occurring on surface-associated nanofabricated metal core–dielectric shell nanostructures. In this arrangement, the nanostructures act as mimics of “free” dielectric nanoparticles in suspension with built-in sensing function. For convenience the used NPS surfaces are described in some detail in the ESI.† The sensor makes it possible to probe molecular adsorption events (kinetics) in situ and in real-time without the risk of sample aggregation or the need for purification procedures. In this way it effectively eliminates some of the key complicating factors of existing corona characterization techniques discussed above. In particular, it enables real-time in situ experiments of both initial corona formation and subsequent evolution. Moreover, this sensing platform is versatile due to the possibility to quite freely tailor the shell material and to fabricate nanostructures of different sizes. In the present work, we use nanoparticles that are comprised of a gold (Au) plasmonic core for molecular detection through localized surface plasmon resonance.18,19 A homogeneous 10 nm SiO2 dielectric layer is grown on the sensor chip surface to encapsulate the Au nanoparticles and in this way form a mimic of silica nanoparticles in solution20,21 (Fig. 1). We also note that the core–shell structures are not perfect spheres but faceted particles. Here, we capitalize on this effect by quantitatively analyzing the role of flat versus curved surfaces on a faceted nanoparticle, in corona formation. Core–shell nanostructures with different flat/curved ratios, applied to probe differences in protein adsorption behavior to the two surface regions, are obtained by fabricating facetted nanostructures of different size. The NPS experiment is different from most previous protein corona studies as the proteins are exposed to the sensor surface under a constant flow. The flow rate may change the binding kinetics and there is some evidence that there is an increased protein adsorption to the surface in a flow compared to a static situation.22–24 Although the applied NPS technique indeed provides relevant new information regarding the corona formation process, the use of multiple, complementary, analytical techniques is of key importance to a thorough characterization of the nanoparticle corona.
![]() | ||
Fig. 2 (A–C) Nanoplasmonic sensing data of the adsorption of bovine serum, IgG depleted bovine serum (bovine serum–IgG), IgG and BSA to the three different types of sensors, that is, 92, 123 and 167 nm average particle size. (D–F) The thicknesses of the formed coronas (after 60 min adsorption) calculated as a function of their refractive indices based on eqn (1). The insets show TEM-images (top view) of the differently sized core/shell nanostructures where the surface facets of the Au-cores are clearly visible (all scale bars equal 20 nm). |
The magnitude of the NPS peak shift is related to both the thickness and the refractive index of the formed corona, according to the well established relation:33
![]() | (1) |
From eqn (1) it becomes clear that it is not possible using nanoplasmonic sensing alone to separate the film thickness of an adsorbed layer and its refractive index, and thus single out their absolute contributions to the experimentally measured peak shift. For this reason, we calculate the thickness of the corona (d) for each sample, using the recorded peak shift after 60 min incubation (Δλ), for a range of refractive indices (nlayer) according to eqn (1). The data are presented in Fig. 2D–F and show that the thicknesses of the acquired coronas vary and that dBSA < dIgG < dbovine serum-IgG ≤ dbovine serum over the entire range of considered refractive indices, assuming a similar refractive index32 of the formed protein coronas.
When comparing the data of the corona formation at the surface of the 92, 123 and 167 nm core/shell nanostructures some interesting differences can be seen. To understand them it is important to note the surface faceting of the nanostructures as shown in Fig. 2. Specifically, this means that it is the ratio between flat and curved areas at the nanoparticle surface that increases with the particle size and not the surface curvature that decreases. In fact the latter remains constant.17
We start our discussion with BSA since we have previously evaluated its response to the size – and thus the flat-to-curved surface area ratio – of the nanostructure,17 using the same experimental approach. From our previous study, in brief, we know that BSA gives rise to a larger plasmon resonance peak shift for the largest nanostructures upon corona formation. In the framework of the model we have developed in that study, this indicates that a denser protein layer is formed at the planar facets compared to the curved regions at the edges between the facets, likely due to increased surface interactions. Interestingly, applying the same concept to the data obtained in the present study, it turns out that the observed effect for BSA is small compared to the other samples, that is, IgG and the two serum samples (Table 1). Specifically, our data show that the plasmonic sensor response for IgG almost doubles relative to the response for BSA when the size of the nanostructures increases from 92 to 167 nm. Similarly, the response for the serum samples relative to the response for BSA also increases with the size of the nanostructures, although not to the same extent as for IgG. From these observations we draw the first conclusion that the corona formation indeed is different for the different systems under study. Specifically, following the same reasoning as developed for BSA in our earlier work,17 the larger fraction of planar areas on the facets of the 167 nm nanostructures compared to the smaller ones, allows for overall increased surface interactions, which in turn generates a globally denser protein corona and thereby a larger plasmonic sensor response. An alternative scenario also explaining the observed size-dependent differences of the plasmonic response to corona formation is that the proteins (especially IgG) adsorb with different orientations depending on where the adsorption occurs, i.e. at flat or curved regions. The importance of surface curvature on the orientation of protein binding to silica nanoparticles has previously been demonstrated for several proteins.28,29 In line with this scenario, IgG forms a thicker corona (e.g. protein adsorbed head on) at the planar facets and a thinner corona (e.g. protein adsorbed side on) at the curved regions. In the complex serum samples, which contain a large variety of different components, a combination of both described scenarios is likely to occur. Thus, due to the significantly larger response obtained for the serum samples, the results indicate that neither BSA nor IgG is a dominating component in the corona formed at the surface of SiO2 nanoparticles in bovine serum.
![]() | ||
Fig. 3 Kinetics of the corona formation upon addition of bovine serum, IgG depleted bovine serum (bovine serum–IgG), IgG and BSA using (A) 92 nm, (B) 123 nm and (C) 167 nm nanoplasmonic sensors. To account for the different sensitivity of the differently sized Au cores the measured peak shifts during corona formation were normalized using the experimentally determined sensitivity factors of the respective sensor (see ESI† for details). Hence, the shown data for the three different nanoparticle sizes can be directly and quantitatively compared. The insets show the normalized rate of adsorption during exposure to the first 0.6 mg of added protein. All experiments were performed under constant flow. For comparison between samples, the accumulated mass of added protein (exposure) is used in the present figure. |
Interestingly, the kinetics for the two serum samples also varies with the size of the nanostructures. For the largest nanostructures the initial rate of adsorption is markedly higher compared to the smallest structures. This result may be explained by (i) a more rapid adsorption to the planar areas compared to the curved regions (higher ratio of flat/curved areas at larger nanostructures), (ii) that flat and curved regions of the faceted nanostructures causes IgG and serum components adsorb with different conformation and/or orientations, or (iii) combinations of (i) and (ii).
From the results in Fig. 4 it is clear that BSA does not form a complete, irreversibly bound monolayer at the surface of the nanostructures, as further adsorption of serum proteins and IgG is possible. These data may be explained by the following two scenarios: (i) BSA adsorption forms an incomplete monolayer, that is, IgG and serum proteins may directly adsorb to the SiO2-surface; (ii) BSA adsorbs reversibly allowing IgG and serum proteins to replace the preadsorbed BSA. Which scenario that is the most likely to occur is discussed in more detail below together with additional data obtained by DLS and SDS-PAGE.
One important difference between the NPS experiments and the experiments in dispersion is that in dispersion we are studying the formed corona in equilibrium, instead of following the corona formation in real time. Four BSA concentrations and three silica particle concentrations were studied in detail. Table 2 shows the radii obtained by DLS for silica particles alone or together with BSA. There is a distinct increase in the size of the BSA incubated silica particles, which corresponds well with a monolayer of BSA. The size of the BSA/silica particle complex is not increasing with increasing concentrations indicating that the particle surface is saturated. However, it should be noted that DLS measures the hydrodynamic radius and that the formation of a BSA layer with low surface coverage cannot be excluded. The corresponding NPS results (Fig. 2) show that the effective refractive index of the BSA corona needs to be small (RI of about 1.36) to obtain a similar thickness of the protein layer, indicating a large degree of hydration.
C (mg ml−1) | Radiusa (nm) | |||||
---|---|---|---|---|---|---|
BSA | 4.9% silica | %Pdb | 3.3% silica | %Pdb | 2.2% silica | %Pdb |
a Radius calculated using a cumulative fit. b % polydispersity. | ||||||
2.1 | 43.9 ± 0.2 | 16 ± 2 | 44.2 ± 0.3 | 13 ± 2 | 43.4 ± 0.6 | 15 ± 1 |
1.4 | 44.3 ± 1.2 | 17 ± 4 | 43.5 ± 0.2 | 15 ± 1 | 43.2 ± 0.2 | 14 ± 2 |
0.9 | 45.0 ± 0 | 16 ± 1 | 43.2 ± 0.3 | 17 ± 1 | 43.8 ± 0.6 | 14 ± 1 |
0.62 | 44.8 ± 0.3 | 16 ± 1 | 42.9 ± 0.1 | 16 ± 1 | 43.4 ± 0.6 | 14 ± 1 |
PBS | 38.3 ± 0.8 | 16 ± 2 | 37.9 ± 0.5 | 14 ± 1 | 38.5 ± 0.7 | 15 ± 2 |
H2O | 37.4 ± 0.2 | 19 ± 1 | 37.8 ± 0.4 | 17 ± 2 | 39.3 ± 0.3 | 17 ± 1 |
Next, we mimicked the sequential corona formation that was performed in the NPS experiments in the DLS analysis. Thus, bare or BSA pre-incubated silica nanoparticles were mixed with bovine serum and after incubation for one hour the size of the protein/particle complexes was measured by DLS. The results show that the radii of the formed protein/particle complexes are similar, regardless if the silica particles were pre-incubated with BSA or not, Table 3. The peak corresponding to the smallest entities likely represent free proteins in the solution. In serum other structures, such as low- and high-density lipoprotein particles, are in the same size range (10 to 100 nm) as the nanoparticles and we cannot distinguish between if the multimodal size distribution arises from only the particles or is a result from the complex environment. However, the size distribution is clearly different from the size distribution of bovine serum alone.34 If the BSA forms an irreversible corona with high coverage the size of the BSA pre-coated particles is not expected to change after mixing with bovine serum. On the other hand, if the coverage is low and/or the adsorbed BSA is exchangeable other serum proteins may still bind, forming a thicker protein layer.
To further investigate whether BSA is irreversibly bound to the nanoparticles SDS-PAGE analysis of the BSA and the sequentially formed coronas was performed. After incubation in BSA or serum the particles were centrifuged and the pellets carefully washed with PBS. The bound proteins were desorbed with SDS and separated by SDS-PAGE. BSA pre-coated particles contain BSA also after the centrifugation and washing steps, Fig. 5A, indicating a strong binding to the silica particles. However, after incubation in bovine serum no BSA is bound to the silica particles, regardless if they were preincubated with BSA or not, see Fig. 5A. The lack of BSA is more easily seen in Fig. 5B where proteins around BSA are more separated. Actually, the protein profile is identical for bare and precoated silica particles, clearly suggesting that adsorbed BSA is exchanged by other serum proteins. This thus corroborates the interpretation of the corresponding NPS data discussed above. The low amount of BSA in the corona was expected, despite its high concentration in serum, as BSA binds with much slower kinetics than other serum proteins as shown in the NPS experiments summarized in Fig. 3 and because the affinity for the surface is likely low compared to other serum proteins. The reversibility of the BSA adsorption to the silica nanoparticles is also demonstrated when BSA precoated particles are incubated with IgG, Fig. 5C. The amount of BSA on the particles is clearly reduced as IgG is binding to the surfaces of the BSA precoated particles. Our results on the effect preadsorption of BSA correspond well with previous results using human albumin and human plasma.35
Finally, using SDS-PAGE, we also demonstrated that a formed BSA corona is largely replaced when the nanoparticle–BSA complex is exposed to bovine serum or IgG, in line with the NPS data. Thus, BSA adsorbed to SiO2 nanoparticles does not prevent further protein adsorption. Similarly, other serum proteins replace IgG when the nanoparticle–IgG complex is exposed to bovine serum. A more detailed summary of the main results, together with a schematic figure thereof, is given as ESI.† In essence, the presented data show the complementarity between the three different analytical techniques applied in this study. Specifically, together they generate time-resolved data of the adsorption processes (kinetics – NPS), data on corona thickness (DLS) and the size/identity (SDS-PAGE) of the adsorbed proteins. Thus, our approach clearly shows that, to better understand processes occurring at the nano–bio interface like corona formation, the use of complementary techniques is essential, and our results open up for further investigations regarding nanoparticle–protein interactions and the process of corona formation in complex biological environments.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6nr06399c |
This journal is © The Royal Society of Chemistry 2017 |