S.
Winzen
a,
S.
Schoettler
a,
G.
Baier
a,
C.
Rosenauer
a,
V.
Mailaender
ab,
K.
Landfester
a and
K.
Mohr
*a
aMax Planck Institute of Polymer Research, Ackermannweg 10, 55128 Mainz, Germany. E-mail: mohr@mpip-mainz.mpg.de
bDept. of Hematology, Oncology and Pneumology, University Medical Center of the Johannes Gutenberg-University Mainz, Langenbeckstr. 1, 55131 Mainz, Germany
First published on 8th January 2015
Here we demonstrate how a complementary analysis of nanocapsule–protein interactions with and without application media allows gaining insights into the so called hard and soft protein corona. We have investigated how both human plasma and individual proteins (human serum albumin (HSA), apolipoprotein A-I (ApoA-I)) adsorb and interact with hydroxyethyl starch (HES) nanocapsules possessing different functionalities. To analyse the hard protein corona we used sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) and a protein quantitation assay. No significant differences were observed with regards to the hard protein corona. For analysis of the soft protein corona we characterized the nanocapsule-protein interaction with isothermal titration calorimetry (ITC) and dynamic light scattering (DLS). DLS and ITC measurements revealed that a high amount of plasma proteins were adsorbed onto the capsules’ surface. Although HSA was not detected in the hard protein corona, ITC measurements indicated the adsorption of an HSA amount similar to plasma with a low binding affinity and reaction heat. In contrast, only small amounts of ApoA-I protein adsorb to the capsules with high binding affinities. Through a comparison of these methods we have identified ApoA-I to be a component of the hard protein corona and HSA as a component of the soft corona. We demonstrate a pronounced difference in the protein corona observed depending on the type of characterization technique applied. As the biological identity of a particle is given by the protein corona it is crucial to use complementary characterization techniques to analyse different aspects of the protein corona.
The newly formed biomolecular interface can be divided into the ‘hard’ and ‘soft’ protein corona, which differ in the binding strength and exchange rates of the proteins associated with the nanomaterial surface. A ‘hard’ protein corona consists of proteins with high binding affinities that are tightly bound and show exchange times greater than the time needed for internalization of a particle.7 This adsorption behaviour is also often referred to as irreversible protein binding. However, proteins which are loosely bound to the nanomaterial surface, or are connected with the hard protein corona via week protein–protein interactions, form the so called ‘soft’ protein corona.3,7–11 These proteins are believed to have high exchange rates and can be replaced easily in a biological environment.7 In practice, the hard corona can be defined as those proteins which are not removed from the nanomaterial's surface during preparation procedures such as washing and centrifugation. These procedures can interrupt the relatively weak protein–protein interactions which would remove the soft protein corona.
However it is still not clear, which form of protein corona exists under physiological conditions that also involve some shear forces in the blood stream. Accordingly it is difficult to determine the biologically relevant entity that for example is read by cellular processing. It has been shown that in some cases the hard protein corona is responsible for the interaction pathways,12 but usually there is no information gained about the role of the soft corona proteins.
The majority of published studies that investigate the protein corona involve isolation of the nanomaterial from the biological media after employing washing steps.8,13–16 As defined above, these investigations tend to explore the hard protein corona. The composition, impact and relevance of the soft protein corona are matters that still require attention.7,17 There are only a few analytical methods available to investigate nanomaterials inside the relevant biological media. Using dynamic light scattering (DLS) one can sensitively detect size changes of particles in undiluted blood plasma.18–20 Additionally, it is possible to monitor the change in heat that results from protein adsorption onto nanomaterials with isothermal titration calorimetry (ITC). As an analytical tool, ITC can provide protein binding affinities and stoichiometry.21,22 An alternative method to investigate nanomaterial–protein interactions in contact media with a high sensitivity is fluorescence correlation spectroscopy (FCS).23–25 The drawback of this technique is that FCS requires either fluorescently labelled nanoparticles or labelled proteins.
Nanomaterials which should be applied as drug delivery devices are often functionalized with poly(ethylene glycol) (PEG) chains to prevent unspecific cell uptake and to suppress protein adsorption to some extent.26–28 Suppressing protein adsorption can increase the circulation time of nanocarriers in the bloodstream, leading to a better bioavailability of transported drugs.26 Since sugar-based polymers like dextranes also exhibit a low protein adsorption,29,30 hydroxyethyl starch (HES) is being discussed as a natural nontoxic alternative for PEG with similar protein repellent characteristics.31–33 The investigated nanocarriers in our study were synthesized from HES to obtain capsules with a shell that decreases protein adsorption.
Here, we have investigated the hard and soft protein corona of HES capsules in human plasma, human serum albumin (HSA) and apolipoprotein A-I (ApoA-I) solution using a combination of sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE), a protein quantitation assay, ITC and DLS. In addition to non-functionalized capsules, we also analysed carboxy-functionalized (HES-COOH) and amino-functionalized (HES-NH2) capsules in order to investigate the effect of different charges at neutral pH value.
This approach enables a comparison between the behaviour of the capsules with plasma proteins and with isolated proteins. HSA was chosen for the analysis because it is highly abundant in blood plasma (∼44 g L−1 average)34 and well characterized. Furthermore, HSA was found to be depleted in the hard protein corona of the previously investigated particles with comparison to its concentration in plasma.13 In contrast, ApoA-I was shown to be enriched in the protein corona of the same nanomaterials as well as liposomes,9,13,28,35 and should be considered an interesting protein for adsorption measurements. Our work effectively shows that different analytical investigations are required to provide a full picture of a nanocapsule's hard and soft protein corona.
Fig. 1 SEM images and characteristics of HES nanocapsules with different surface functionalizations. |
All nanocapsules exhibited a negative zeta potential due to a redispersion in water with sodium dodecyl sulphate (SDS). The capsules are purified by extensive dialysis after redispersion to remove most of the SDS. However a small amount is needed on the surface to keep the capsules stable in solution. A functionalization with NH2-groups on the capsule surface leads to a slight increase in zeta potential because of their positive charge at neutral pH. The amount of functional groups per nanocapsule was determined using particle charge detection (PCD, see Methods section).
Fig. 2 Coomassie-stained gel after SDS-PAGE of plasma proteins recovered from the surface of differently functionalized HES nanocapsules. Pure plasma (right lane) served as a reference. |
A quantitative analysis of the adsorbed proteins was performed via a Pierce 660 nm protein assay. Samples were prepared following the SDS-PAGE procedure and analysed photometrically. The obtained protein amounts were normalized with regards to the nanocapsules’ surface area and are shown in Fig. S1.† For the non- and amino-functionalized capsules, values of 0.47 ± 0.11 mg m−2 and 0.46 ± 0.09 mg m−2 respectively were obtained, whereas for the carboxy-functionalized capsule the value decreased to 0.32 ± 0.01 mg m−2. This quantitation suggests that the carboxy-functionalized capsules bind fewer proteins than the other two types of capsules. The described methods require extraction from the interaction medium before analysis. Consequently, proteins that possess a binding affinity below a certain threshold, currently undefined, are likely to be removed during sample preparation. Therefore, we further applied ITC and DLS to investigate the protein–nanocapsule interaction directly in the application medium.
Capsule | Protein | ΔH/kJ mol−1 | K a/106 L mol−1 | ΔS/J K−1 mol−1 | N | Number of proteins per 10 nm2 |
---|---|---|---|---|---|---|
HES | Plasma | −285 ± 97 | 1.0 ± 0.4 | −845 ± 329 | 182000 ± 39000 | 2.00 |
HSA | −277 ± 43 | 0.8 ± 0.3 | −818 ± 147 | 114000 ± 16000 | 1.25 | |
ApoA-I | −6010 ± 185 | 333 ± 124 | −20000 ± 613 | 10 ± 4 | 1.05 × 10−4 | |
HES-COOH | Plasma | −306 ± 192 | 1.0 ± 0.8 | −1217 ± 318 | 70000 ± 13500 | 1.40 |
HSA | −308 ± 30 | 0.4 ± 0.2 | −928 ± 103 | 156000 ± 36000 | 3.00 | |
ApoA-I | −5150 ± 787 | 188 ± 79 | −17100 ± 2640 | 6 ± 3 | 1.19 × 10−4 | |
HES-NH2 | Plasma | −281 ± 139 | 1.2 ± 0.6 | −827 ± 468 | 121000 ± 29000 | 1.40 |
HSA | −277 ± 45 | 0.5 ± 0.1 | −820 ± 152 | 184000 ± 33000 | 2.00 | |
ApoA-I | 883000 ± 24000 | 5.4 ± 0.1 | 2.4 ± 0.9 × 106 | 8 ± 6 | 0.97 × 10−4 |
These experiments determined that the interactions between human plasma and each type of HES capsule are exothermic and therefore enthalpically favoured. For the analysis of the plasma measurements, the molarity refers to the average concentration of HSA in the plasma. Increasing the concentration consequently leads to a slightly higher stoichiometry N, while the other parameters do not change significantly. The results obtained from the fitting (see Table 1) are in agreement with the analysis of the hard protein corona. The enthalpy change is similar for each different surface functionalization. Furthermore, Ka and ΔS are the same. Given the high number of proteins interacting with one capsule in combination with the small reaction enthalpy, it can be concluded that the proteins are loosely associated with the capsule surface. Additionally, the number of proteins per 10 nm2 of surface area was calculated based on the stoichiometry and the hydrodynamic radius of the capsules. For all functionalities a similar number of around 1–2 proteins was obtained. Thus, the lower protein amount on the carboxy-functionalized particles measured with the protein assay is due to preparation effects. It is important to note that 1–2 proteins per 10 nm2 is not enough space for one protein to obtain a monolayer spread out flat on the surface. Groups of proteins are likely adsorbing at the same time, e.g. protein–protein aggregates formed previously in solution.
The adsorption of pure HSA onto the HES capsules was investigated. In general, the titration isotherms and fit parameters obtained from plasma and HSA are quite similar. This can be explained by the high HSA content in human plasma. Still, slight deviations of Ka, N and the protein per surface area values can be observed. For all capsule types, Ka of HSA was reduced compared to the Ka of plasma, which is the result of additional proteins in the plasma that have a higher binding affinity than HSA. The number of proteins per surface area changes only slightly in comparison to plasma. Those changes can be attributed to different protein compositions of the soft coronas formed in plasma. Moreover, the affinity of HSA towards the HES capsules is roughly 4 times smaller than the protein's affinity towards polystyrene particles, which has been previously investigated.40 This finding can be explained by taking into account the more hydrophilic nature of HES compared to polystyrene, which results in less hydrophobic interactions between the capsule surface and the proteins. However, the number of proteins per 10 nm2 is approximately 10–20 times higher than the values reported for the polystyrene particles.40 From the protein amounts adsorbed it can be concluded that the soft corona formed for HES nanocapsules contains much more proteins than the soft corona of polystyrene particles. Therefore, the difference between the hard and soft corona is more significant for the HES capsules.
The low affinity of HSA to the HES capsules, given by the parameter Ka, in combination with the stoichiometry and the depletion of HSA in the hard protein corona (observed in SDS-PAGE experiments) leads to the identification of HSA as a soft corona protein. These results match the findings reported in literature.41,42
In contrast to the HSA adsorption, a significant amount of heat is generated when ApoA-I is titrated into pure HES capsules (see Fig. 3). Also, Ka differs by about an order of magnitude, which implies that the capsules have a greater affinity towards ApoA-I than to HSA. Meanwhile, N and the protein number per surface area are several orders of magnitude smaller than the values for the HSA molecules. These findings identify ApoA-I as a hard corona protein for the investigated system. This is in good agreement with a previous study of Cedervall et al., who found a high binding affinity of ApoA-I to N-isopropylacrylamide (NIPAM) particles.43 Our results also support the theory that the high binding affinity of ApoA-I is due to the surface characteristics of the nanomaterial rather than the adsorption of fats from the plasma before interaction with the protein.
While the non-functionalized and the carboxy-functionalized capsules display similar behaviour, the amino-functionalized capsules maintain an endothermic interaction with ApoA-I. The large amount of heat absorbed during the titration suggests an entropy-driven interaction process. The independent binding fit (Table 1) confirms these findings, revealing an entropy gain contrasting the entropy loss in the other interactions studied. This entropy gain could partially be attributed to protein structural changes and unfolding during adsorption. An unfolded protein occupies a higher surface area on the capsule and, therefore, more water molecules of the hydration shell are released. However, the amount of heat generally needed for an unfolding of proteins44,45 is lower than the heat absorbed in this reaction. This suggests that there is another interaction process going on, which could involve the surfactant SDS. To clarify this, we are continuing to investigate when denaturation occurs and to which extent surfactants influence the adsorption processes. At least the surfactant does not seem to screen the functional groups of the capsule surfaces, since the endothermic reaction was only observed for the NH2-functionalised capsule.
(1) |
(2) |
Knowing the ACF of human plasma and the respective HES capsule, the correlation function of the plasma capsule mixtures could be analysed. If no aggregation were to occur, the resulting ACF of the plasma capsule mixture would correlate to the so-called force fit. In the force fit, the sum of the individual correlation functions with the known parameters of the two compounds (plasma/capsule) is kept fixed and the intensity contributions for plasma fP and capsule fC are the only fit parameters (eqn (3)).
g1,m(t) = fPg1,P(t) + fCg1,C(t) | (3) |
However, interactions of HES capsules with plasma components resulted in larger sizes than the plasma components and HES capsules themselves; consequently, the ACFs could not be described by the force fit.
The fit needed to be modified by an additional, longer ACF relaxation time related to the size of the formed structures (eqn (4)).
g1,m(t) = fPg1,P(t) + fCg1,C(t) + faggg1,agg(t) | (4) |
(5) |
The multicomponent analyses of the HSA and ApoA-I capsule mixtures were performed accordingly. The exemplary analysis of the ACF (g1,m(t)) of the mixture of non-functionalized HES capsules with human plasma is shown in Fig. 4A. The multicomponent analyses of the other mixtures are provided in the ESI.† The force fit (sum of individual components) did not describe the data correctly. In contrast, the fit including an additional aggregate component was suitable for the capsule/plasma mixture. The fitting procedure for each angle produced the corresponding hydrodynamic radius Rh shown in Fig. 4B. From the extrapolated diffusion coefficients the z-averaged hydrodynamic radius <1/Rh>z−1 was calculated by the application of Stokes law and can be found in Table 2. Fig. 4 demonstrates the increase of Rh, which occurs after interaction with plasma. The size increase of around 100 nm for non-functionalized capsules is too small to be caused by aggregation between several capsules, so it is attributed to coating with plasma proteins. Additionally, the intensity contribution of the aggregates (see Table 2) was calculated and is also shown in Fig. 4B. The fraction of the new species (protein-coated capsules) being formed is significant in comparison to the fraction of pure capsules. The remaining fraction of the capsules in the mixture can be attributed to the natural distribution of the thickness of the protein corona and the polydispersity of capsules and aggregates. As the applied method is highly sensitive for the detection of aggregates with sizes larger than the largest size present in the pure components solutions, changes within the size distribution of plasma and capsules are only detectable if the amplitudes (i.e. the intensity fractions) of the newly formed sizes are sufficiently large (and thus detectable by DLS). Typically, intensity fractions between 3% (for sizes larger than the largest component in the mixture) and 20% (for sizes in the same size range of the mixture components) of newly formed particles are necessary in order to become detectable by the described fitting procedure.48 Consequently, small capsules with a thin protein corona are not recognized as aggregates. The DLS analysis has been carried out for all capsule types and proteins and the corresponding sizes and intensity fractions are displayed in Table 2. The DLS experiments were repeated after 24 h of incubation time to check for any changes related to the protein adsorption kinetics. However there was no significant difference found in the aggregation behaviour. According to reports in the literature, the composition of the protein corona changes quantitatively but not qualitatively. Therefore it is likely that slight changes in the protein corona compositions of the mixtures with plasma cannot be detected with this type of analysis.
Capsule | No protein | Plasma | HSA | ApoA-I | ||||||
---|---|---|---|---|---|---|---|---|---|---|
R h/nm | ξ-potentiala/mV | R h,Agg/nm | I%Aggb | ξ-potentiala/mV | R h,Agg/nm | I%Aggb | ξ-potentiala/mV | R h,Agg/nm | ξ-potentiala/mV | |
a ξ-potential at pH 7 in 0.001 M KCl solution. b Intensity fractions of the aggregates (I%Agg) are exemplarily given for a scattering angle of 64°. For the intensity fractions of other scattering angles see Fig. S3, S5 and S6. | ||||||||||
HES | 275 ± 26 | −34 ± 3 | 390 ± 39 | 33 | −22 ± 9 | 347 ± 35 | 28 | −18 ± 1 | — | −7 ± 6 |
HES-COOH | 200 ± 20 | −34 ± 3 | 319 ± 32 | 25 | −13 ± 5 | 352 ± 35 | 47 | −15 ± 4 | — | −8 ± 7 |
HES-NH2 | 256 ± 26 | −27 ± 3 | — | — | −11 ± 4 | — | — | −10 ± 1 | — | −18 ± 8 |
Additionally, the zeta-potential was measured for all samples before and after the mixture with proteins (see Table 2) to monitor the change of the capsules’ surface charge after protein adsorption.
The average size increase of around 120 nm for the non- and carboxy-functionalized capsules coated with plasma (Fig. S5†) and the significant intensity fraction of the resulting larger structures can be attributed to the formation of a protein corona. For the amino-functionalized capsules no component larger than the components of the mixture was found via DLS (Fig. S8†), even though the ITC measurements suggested adsorption processes similar to the other capsules. The DLS results are in agreement with previous studies of amino-functionalized nanoparticles that did not show any aggregate formation.20 It has to be noted that the aggregation behaviour cannot be attributed to a positive overall particle charge due to the amino functionalization as the amino-functionalized particles observed a negative zeta potential because of their redispersion in water with SDS. Concerning the difference between the findings from ITC and DLS, it is important to note that the samples taken from ITC measurements were diluted for the DLS experiments (see Methods section). While a high concentration is needed to obtain a sufficient signal in ITC, it has to be decreased for light scattering due to the high turbidity of the capsules above a certain concentration. Light scattering measurements would then be complicated by multiple and back scattering artifacts.49,50 Therefore, it is possible that a formed protein corona changes upon dilution due to concentration dependencies. The soft protein corona is especially affected because of the high protein exchange rates reported in literature.7 Nevertheless, the zeta-potential for all three surface functionalizations was increased after protein adsorption. This suggests coverage with proteins, as the negative charges from SDS are shielded, or even replaced, by proteins with a lower negative charge compared to their volume or positive charge.
The same analysis was performed with HSA and yielded results similar to the plasma measurements (see Table 2, Fig. S3 and S6†). This reaffirms the conclusion drawn from the ITC experiments that HSA is adsorbed to the capsule surface with a high stoichiometry and thus leads to the formation of a thick soft protein corona. The combination of these results with the depletion of HSA in the SDS-PAGE identifies HSA as a soft corona protein. Again, no additional aggregates were found in the sample containing amino-functionalized capsules (Fig. S9†). Also, the zeta-potential increase suggests the coverage with proteins.
In contrast to plasma and HSA, no additional aggregates were formed from the interaction of all capsule types with ApoA-I (see Fig. S4, S7 and S10†). This agrees with the ITC results, given that few molecules (around 10 ApoA-I molecules per capsule) adsorbed to the capsule surface. The formation of aggregates was not expected because the size increase due to formation of an ApoA-I monolayer would not be greater than 8 nm in radius. The size of the protein (Rh ≈ 3.8 nm)51 is only around 1.5% of the capsule size, so the size change is in the experimental error of the light scattering experiment. These results in combination with the high binding affinity for ApoA-I (Table 1) classify ApoA-I as a hard corona protein.
Blood was taken at the Transfusion Centre of the University Medical Centre of the Johannes Gutenberg-University Mainz from 10 healthy donors after obtaining informed consent. The study was approved by the local ethics committee. To prevent blood clotting Li-Heparin was added. The blood was centrifuged to pellet red and white blood cells and the plasma supernatant was pooled. Aliquots were stored at −80 °C. After thawing, the plasma was centrifuged at 20000g for 1 h at 4 °C to remove any residual protein precipitates. A protein concentration of 66 g L−1 was determined for the plasma.
For measurements of nanocapsules–protein mixtures the samples were prepared according to the ITC titration procedure. In every case, 300 μL of a capsule suspension was mixed with 50 μL of protein solution or water for the reference measurements (concentrations were the same as in ITC experiments). Plasma and individual proteins were prepared by adding the solutions into 300 μL of water to maintain the same dilutions. All plasma and HSA containing samples were then diluted with water up to 2 mL sample volume. ApoA-I containing samples were diluted up to 5 mL total volume. Capsule concentrations of 0.015 g L−1 and 0.24 g L−1 respectively were achieved. The solutions were then filtered through Millex SV filters with a pore size of 5 μm (Merck Millipore, Billerica, USA) into dust-free quartz light scattering cuvettes (inner diameter 18 mm, Hellma, Müllheim), which were cleaned before with acetone in a Thurmont-apparatus.
Footnote |
† Electronic supplementary information (ESI) available: Pierce 660 nm protein assay, ITC data evaluation, DLS data evaluation, autocorrelation functions of protein – HES capsule mixtures. See DOI: 10.1039/c4nr05982d |
This journal is © The Royal Society of Chemistry 2015 |