Maria
Kokkinopoulou‡
a,
Johanna
Simon‡
a,
Katharina
Landfester
a,
Volker
Mailänder
*b and
Ingo
Lieberwirth
*a
aMax Planck-Institute for Polymer Research, Ackermannweg 10, 55128 Mainz, Germany. E-mail: lieberw@mpip-mainz.mpg.de
bClinic of Dermatology, University Medicine Mainz, Langenbeckstraße 1, 55131 Mainz, Germany. E-mail: volker.mailaender@unimedizin-mainz.de
First published on 13th June 2017
The use of nanocarriers in biology and medicine is complicated by the current need to understand how nanoparticles interact in complex biological surroundings. When nanocarriers come into contact with serum, proteins immediately adsorb onto their surface, forming a protein corona which defines their biological identity. Although the composition of the protein corona has been widely determined by proteomics, its morphology still remains unclear. In this study we show for the first time the morphology of the protein corona using transmission electron microscopy. We are able to demonstrate that the protein corona is not, as commonly supposed, a dense, layered shell coating the nanoparticle, but an undefined, loose network of proteins. Additionally, we are now able to visualize and discriminate between the soft and hard corona using centrifugation-based separation techniques together with proteomic characterization. The protein composition of the ∼15 nm hard corona strongly depends on the surface chemistry of the respective nanomaterial, thus further affecting cellular uptake and intracellular trafficking. Large diameter protein corona resulting from pre-incubation with soft corona or Apo-A1 inhibits cellular uptake, confirming the stealth-effect mechanism. In summary, the knowledge on protein corona formation, composition and morphology is essential to design therapeutic effective nanoparticle systems.
The proteomic composition,10 size and aggregation11 effects of the protein corona are well-known, but its morphology has still to be examined. The protein corona is usually shown either as a uniform layer or as multiple layers covering the nanoparticle. For the first time we have been able to visualize the morphology of the protein corona and provide a 3D model of its structure by using transmission electron microscopy. In contrast to existing theories and sketches, we found that the protein corona forms a loose network which is attached to the nanoparticle.
In this paper we focus on three different polystyrene nanoparticles (plain, carboxyl-functionalized and amino-functionalized). These particles are easily synthesized in a wide range of sizes/surface functionalization and are ideal candidates for studying bio–nano interactions.12 The protein corona morphology and composition of those nanoparticles were compared using TEM, DLS and LC-MS. Further we carefully monitored the process from initial corona formation directly after incubation in human serum and after each washing/centrifugation steps. Using a 3D model reconstruction, we were able to quantify the amount of adsorbed protein covering the nanoparticles. Finally, endocytosis and intracellular trafficking of the nanoparticle coated with or without protein corona was investigated after incubation with macrophages. These studies offer a better understanding of the biological identity of the nanoparticles and will therefore contribute to a safer and more effective application in nanomedicine.
Complementary analytical methods were applied to visualize the structure (TEM) of the protein corona and determine changes in size. The methods included multi-angle DLS as well as characterizing its composition by label-free, ultra-pressure liquid chromatography mass spectrometry (UPLC-MS).
The morphology of the protein corona was studied by TEM using a negative staining technique.17 Proteins were embedded in a thin, free standing layer of dried trehalose containing heavy metal salts (e.g. uranyl acetate) providing high contrast samples, suitable for conventional electron tomography.18,19 In order to assure that the corona morphology is not affected by the embedding into the trehalose film, additional cryo-TEM examinations have been performed. This showed no noticeable structural difference between cryo-TEM and trehalose embedding preparation (Fig. S1†). The protein corona of nanoparticles could be visualized for the first time using this technique and subsequently quantitatively analyzed using a 3-D reconstruction model.
Directly after incubation of nanoparticles in human serum (Fig. 1A), un-functionalized polystyrene nanoparticles (PS) were surrounded by a protein cloud, which appeared to be larger than the diameter of the nanoparticle. Given the average diameter of 140 nm for un-functionalized polystyrene nanoparticles (ESI Tables 1/2†) the average additional corona was estimated to be ∼70–100 nm thick (Fig. 1A) and referred to as the soft protein corona. Interestingly, the soft protein corona is not shown by TEM to be as uniformly distributed and well-rounded, but rather as an undefined network surrounding the nanoparticle. Generally2,20 it is described as highly dynamic layer of proteins which have high exchange rates and low binding affinities towards the nanoparticle. By adding trehalose and uranyl acetate (UA) for TEM sample preparation, the highly dynamic structure is fixed. At this stage a “snapshot” of the nanoparticles surrounded by the soft corona was taken, revealing interestingly its non-uniform structure. In addition, cryo-TEM examinations (ESI Fig. 1†) corroborated the morphology of the protein corona observed by the trehalose-UA preparation. As a result structural artifacts caused by negative staining can be excluded. While there is extensive literature dealing with the subject of hard protein corona, only limited analytic methods used to study the soft protein corona are available.21 Consequently, determining the biological relevance of the soft corona has been slowed down.22 Hence, for in vivo application of nanocarriers it is essential to concentrate on studying the interactions of nanoparticles within a given protein source (serum, plasma) and to analyze their aggregation behavior in such complex surroundings as this can highly affect the biodistribution.11 By using multi-angle dynamic light scattering (DLS) it is possible to measure the average, hydrodynamic size of nanoparticles directly incubated in the protein source.23 Multi-angle DLS was performed at 37 °C in human serum and we observed an average size increase of the hydrodynamic radius of about ∼70 nm for functionalized PS-NPs (ESI Table 2†). This size increase can be attributed to the protein corona formation and correlates well with the visualization of the protein corona in TEM images (Fig. 1A). The data evaluation procedure (ESI Material/methods†) and the auto-correlation functions respectively shown for scattering angle of 30° can be found in the ESI Fig. 5–7.†
In order to differentiate between the loosely bound proteins forming the soft corona and those which can be assigned to the hard corona, we performed an incremental separation process. After incubation with the protein source, the PS-NPs were (i) centrifuged and the supernatant was removed and (ii) afterwards the PS-NPs were washed. The centrifugation and washing was repeated 3 times and at every step the protein corona was characterized.
As previously mentioned, in most studies dealing with the protein corona only proteins of the hard corona have been analyzed. These are tightly bound and have a high affinity towards nanoparticles.24 The hard protein corona of nanoparticles is usually isolated using multiple centrifugation and washing steps to remove loosely or unbound proteins from nanoparticles surrounded by the hard corona.25 Several other studies have compared the protein pattern obtained after centrifugation with different preparation techniques (e.g. magnetic separation, gel-filtration) and these studies have demonstrated that the general protein adsorption pattern is comparable.26,27
Our goal was to analyze the evolution of the protein corona formed directly after incubation in human serum and after each purification/washing step by isolating protein coated nanoparticles via centrifugation. We monitored the morphological development of the protein corona for un-functionalized (Fig. 1) and functionalized nanoparticles (Fig. 3A) by electron microscopy and analyzed the adsorption pattern quantitatively using LC-MS.
After the first centrifugation, un-functionalized PS nanoparticles were surrounded by a substantial protein cloud (Fig. 1B, C and ESI Fig. 2†) as observed directly after incubation (Fig. 1A). There is no significant morphological difference between the protein corona formed directly after incubation and after the first centrifugation step. Only that most of the unbound proteins that were present in the first case were washed away. The 3D reconstructions of the washed samples demonstrate the heterogeneity of the protein corona at this step (Fig. 1C). However, it is difficult to distinguish between tightly or loosely bound proteins associated to one nanoparticle and to unbound proteins. TEM images and DLS measurements reveal a high quantity of proteins adsorbing onto nanoparticles directly after incubation (Fig. 2C). This results in a marked increase in the size of the nanoparticles due to the formation of the soft corona. The soft corona was then washed off after several centrifugation steps and the nanoparticle–particle complex with tightly bound proteins (thus forming a hard corona) could then be isolated (Fig. 1D–G). This result is in agreement with the quantity of proteins quantified after each washing step which continuously decreases with washing steps (Fig. 2D). The number of loosely and unbound proteins was significantly reduced after the first wash (1st centrifugation = 20.02 ± 1.00 mg vs. 1st wash = 2.30 ± 0.10 mg m−2 un-functionalized PS-NP). After the second and third washing steps, the quantity of adsorbed proteins did not decrease significantly. After the third and last wash the protein amount in the supernatant was below the detection limit indicating that the loosely/unbound proteins were effectively washed away (ESI Table 4†). Using the 3D reconstruction presented (Fig. 1G) the number of proteins was counted every 10 nm in the z-axis (assuming that the average protein size is 10 nm). After the first centrifugation, un-functionalized nanoparticles were surrounded by ∼1200 proteins. This number drastically decreased down to ∼400 proteins after the final washing step (Fig. 2C).
There were no obvious structural differences observed in the nanoparticles–protein complexes when washed just once or three times (Fig. 1D–F). In addition, the ζ-potential of nanoparticles (−3.7 mV, measured in KCl) dramatically decreased to −21 mV after incubation in human serum, indicating protein adsorption and corona formation. Again, there were no major changes measured after the first, second and third washing steps (ESI Fig. 9†). After the third and last wash, the corona thickness measured approximately 15 nm (Fig. 1F and G) by TEM. The proteins were tightly bound to the PS nanoparticles and covered most of each periphery. In order to determine statistical robustness, the radius of more than 20 particles from the TEM micrographs was measured (Fig. 1F and ESI Fig. 3†). The hard corona radius measured by DLS was comparable (19 ± 2 nm, ESI Fig. 8†).
In order to monitor the evolution of the hard protein corona, we analyzed the protein composition of un-functionalized PS-NPs after each purification step. A detailed list of all identified proteins is found in the ESI Tables 7 and 8.† The protein composition of the protein corona is significantly altered when compared to the composition of proteins in serum.14,15 The relative amount of abundant serum proteins (>1%) was found to be drastically decreased in the case of human serum albumin, serotransferrin, haptoglobulin, hemopexin and complement C3 after the first centrifugation stage. All these proteins were detected in small quantities in the hard corona (Fig. 2B). In contrast, there was an enrichment of low abundance proteins (clusterin, apolipoprotein A1 and antithrombin-III) which are referred to as the hard corona proteins (Fig. 2B). Significantly, we were able to show that the relative abundance of the identified proteins remained stable after the first washing (Fig. 2A/B). Only slight changes in the protein pattern could be observed after the second and third washing steps. By doing this the equilibrium between proteins in solution and nanoparticle bound proteins changed rapidly and nanoparticles with tightly-bound hard corona proteins could be isolated. These results highlight the crucial purification step (after centrifugation 1 → wash 1) as here the equilibrium between unbound and nanoparticle bound proteins is critically shifted. It should be mentioned that the hard corona is stable and an exposure to new environments could only lead to partial replacement of the proteins. Since these molecules stay for longer times than the characteristic timescale of a given biological process, their presence is relevant for the biological identity of the nanoparticles.28,29 The soft corona on the other hand could be exchanged within seconds. These interactions are crucial and it should not be forgotten though that it is challenging to separate the two in a biological context.30 But in most cases the time scales for biology match the washing concepts, as in the case of particle internalization that could take place in minutes. The concept of these two time scale (hard and soft corona) in protein kinetics is described in more details elsewhere.31
In addition, it is known that the protein adsorption pattern is highly influenced by surface functionalization of nanoparticles.24,32,33 We found that the absolute number of bound proteins per defined surface of nanoparticles is significantly higher (Fig. 3C) for negatively charged nanoparticles (PS-COOH) in comparison to un-functionalized (PS), or positively charged nanoparticles (PS-NH2). In addition, vitronectin was highly enriched in the protein corona of PS-COOH (Fig. 3A) and PS-NH2 particles specifically adsorbed clusterin (ESI† detailed protein list). Clusterin was also found to be the major hard corona protein of poly(phosphoester)- and PEGylated modified PS-NPs.33 The set of PS-NPs in this experiment was stabilized with Lutensol AT-50 which is a PEG-analog surfactant. In addition, there are specific proteins (immunoglobulin k, immunoglobulin γ or apolipoprotein AI) which were identified on all polystyrene nanoparticles under investigation (protein classification ESI Fig. 11†). Previous studies14,15,34,35 support our findings as they recorded an enrichment of apolipoproteins in the hard corona of polystyrene nanoparticles.
After the first centrifugation, at a first glance, there were no major differences (SDS-PAGE, Fig. 3B) between the identified proteins with regard to the surface functionalization of the nanoparticle (PS, PS-COOH, PS-NH2). Additionally, the protein pattern is comparable to the proteins identified in human serum. The major protein band (∼62 kDa) is referred to as human serum albumin as it is the most abundant serum protein. At this stage it was not possible to clearly separate nanoparticles with tightly surrounded proteins from the remaining proteins in solution because their protein concentration was much higher.
Interestingly, some distinct bands were detected in the SDS-PAGE (marked by a red star, Fig. 3B) even without performing any washing steps. These bands were identified as low abundance proteins in a detailed proteomic analysis by LC-MS (e.g. vitronectin, clusterin, apolipoprotein A1). The concentration of low abundance proteins is much higher on the NP than in the pristine human serum. These proteins were further enriched in subsequent washing steps and identified as the proteins of the hard corona (Fig. 3A). There were no significant structural differences (Fig. 3A) visualized in the soft and hard protein corona of functionalized nanoparticles. Additionally the hard corona radius (ESI Table 3†) measured by TEM and DLS was comparable to un-functionalized nanoparticles (10–20 nm).
To visualize the binding of individual hard corona proteins, un-functionalized nanoparticles were incubated in the respective isolated proteins (clusterin, apolipoprotein A1 or IgG) for 1 h at 37 °C and the protein adsorption was analyzed by TEM (Fig. 5 and ESI Fig. 4†). Micrographs show that in most cases the protein was present around the entire periphery of the nanoparticles. In addition, the corona diameter was quite large (∼100 nm), as seen in the case of apolipoprotein A1 (Fig. 5A). However, the secondary structure of the proteins can be altered due to the adsorption on a surface36,37 further determining the cellular uptake.38,39
Therefore, we further studied the interactions of protein corona coated nanoparticles (Fig. 4/5) and a macrophages cell line, RAW264.7. Flow cytometry analysis, confocal laser scanning microscopy (CLSM) and TEM revealed the intracellular uptake of BODIPY-labeled PS-NPs and no adherence to the plasma membrane (Fig. 4B, ESI Fig. 14†). Further, we investigated the difference between nanoparticles surrounded by the hard corona vs. soft corona. Hard corona coated nanoparticles incubated in serum were isolated via repetitive centrifugation and washing (3 times, as prepared for protein corona analysis) and added to serum free cell culture medium (+ hard corona). Additionally, nanoparticles were directly added to serum free culture medium (− corona) or cells cultured in 100% serum (++ soft corona).
Cellular uptake of soft corona coated nanoparticles (++) was strongly reduced in comparison to hard corona coated (+) or uncoated nanoparticles (−). Here, it was shown that the surface functionalization highly influenced uptake behavior (Fig. 4A). Un-functionalized and amino-functionalized nanoparticles were taken up to a significantly lower extent compared to carboxy-functionalized nanoparticles. Interestingly, this trend was observed with or without protein corona (Fig. 4A, ESI Fig. 15†). Additionally, it was found that cellular uptake of PS-COOH nanoparticles was significantly enhanced for nanoparticles surrounded by hard corona proteins compared to uncoated PS-COOH (p < 0.001 ***; ESI Fig. 14†). This is remarkable as we demonstrate that immunoglobulins are the major protein corona component (about 50%) for all investigated nanoparticles. As the morphology of the protein corona was visualized as an undefined loose network of proteins, this indicates that next to the presence of the protein corona the underlying surface functionalization affects cellular uptake.
Transmission electron microscopy (Fig. 4B) was used to compare the internalization mechanism and intracellular trafficking of un-functionalized nanoparticles without (−) or with hard protein corona (+). Numerous un-functionalized polystyrene nanoparticles were packed in long membrane structures (Fig. 4B 2, 3) that resemble the uncoated carriers or early endosomes present in the CLIC/GEEC endocytosis pathway.40 In the presence of protein corona, these structures were not observed. Instead, un-functionalized PS-NPs (either individually or in a group of 2–3) were packed in small vesicles, and were not found inside endosomes (Fig. 4B 5, 6). These results indicate that in the presence of protein corona, the cells choose an alternative endocytosis pathway. Additional TEM micrographs are summarized in the ESI Fig. 16/17.† The exact mechanism will be further studied with the use of antibodies and it is not further discussed here as it is beyond the scope of this paper.
Several studies have shown that the secondary structure of the proteins can be altered by adsorption on a surface36,37 and thus further determine cellular uptake.38,39 As seen in the case of apolipoprotein AI (ApoAI), we found that the corona diameter after incubation with un-functionalized PS-NP was quite large (∼100 nm, Fig. 5A). Therefore, we studied whether this affects the cellular uptake behavior. Un-functionalized nanoparticles were incubated with ApoAI and cellular uptake was analyzed by flow cytometry (Fig. 5B), cLSM (Fig. 5C), and TEM (Fig. 5D). It was found that due to pre-coating with ApoAI cellular uptake was strongly inhibited compared to uncoated (−) or hard corona coated nanoparticles (+), hence it was comparable to soft corona coated nanoparticle (++). Interestingly, the structural properties of the ApoAI and soft protein corona indicated strikingly similar properties (Fig. 5Avs.2C average corona diameter and number of corona proteins).
Fig. 5 A. TEM micrographs and 3D reconstruction images of un-functionalized PS-NP (0.05 m2) incubated with ApoAI1 (100 μg) for 1 h, 37 °C. The additional corona diameter (∼100 nm) and number of ApoAI proteins (>1400 proteins) surrounding un-functionalized PS-NP is comparable to the soft corona images (Fig. 1 and 2). Scale bar: 100 nm. B. Flow cytometry analysis: RAW264.7 cells were incubated with 300 μg mL−1 of un-functionalized (−), protein corona coated (+/++) or ApoAI coated un-functionalized PS-NP for 1 h. C. Confocal laser scanning microscopy images of ApoAI or hard corona coated un-functionalized nanoparticles. The cell membrane is stained using CellmaskOrange (pseudo-coloured red) and BODIPY labeled nanoparticles are pseudo-coloured green. Scale bar: 10 μm. D. TEM micrographs of high-pressure frozen macrophages treated with 300 μg mL−1 of un-functionalized nanoparticles pre-coated with ApoAI indicating a strongly reduced uptake of nanoparticles. Scale bar: 1 μm. |
Electron micrographs were taken on an Ultrascan 1000 (Gatan) charge-coupled device (CCD) camera. The TEM was operated at 200 kV. The Digital Micrograph software (Gatan) was used to collect the images.
In order to view the protein corona in 3D, tilt series over a tilt range of −65° to +65° were recorded at a magnification of 22000×. The SerialEM software (Mastronarde, 2005) was used to collect the tilt series.
The diameter of the particles and the protein corona that was formed around them was calculated with ImageJ.
The ACF of human serum is approximated fitted by a sum of three exponential terms as given in eqn (S1):
(S1) |
(S2) |
If nanoparticles are exposed to human serum the combination of the ACFs of serum and nanoparticles is analyzed. If no aggregation occurs, the resulting ACF of the mixture can be fitted by the sum of two individual ACFs g1,m(t), the so named forced fit(S3).
g1,m(t) = fPg1,P(t) + fNPg1,NP(t) | (S3) |
If aggregation formation of nanoparticles in serum occurs, the ACF cannot be describe by sum of two components, so an additional term ACF g1,agg(t) for aggregates is needed (S4).
(S4) |
Therefore, the correlation function g1,m(t) consists of three terms including the aggregation terms with the intensity contribution fagg.
g1,m(t) = fPg1,P(t) + fNPg1,NP(t) + faggg1,agg(t) | (S5) |
The hard protein corona surrounding nanoparticles was isolated via centrifugation and loosely or unbound proteins were removed. Nanoparticles were centrifuged for 1 h, 20000g (4 °C). The supernatant was collected for protein quantification. The remaining nanoparticle pellet was either re-suspended in water (final nanoparticle concentration of 10 mg mL−1) and analyzed by TEM and DLS or washed with 1 mL of water. This procedure was repeated three times to ensure that all unbound proteins are removed. To elute bound proteins form the nanoparticles, the pellet was re-suspended in 2% SDS (62.5 mM Tris·HCl), heated up to 95 °C for 5 min and centrifuged for 1 h, 20000g (4 °C). The remaining supernatant was collected further analyzed by Pierce Assay, SDS-PAGE and LC-MS.
A detailed overview of all identified proteins is found in separate ESI† (excel sheet).
Hard protein corona coated nanoparticles were prepared as described above (Protein corona preparation). Therefore, the nanoparticles were first incubated with human serum, centrifuged and washed to remove unbound proteins. Protein coated nanoparticles were re-suspended in serum free medium (final concentration: 300 μg mL−1) and incubated with the cells for 1 h.
For flow cytometry experiments, adherent cells were washed with PBS and detached from the culture vessel with 2.5% trypsin (Gibco, Germany) and measurements were performed on a CyFlow ML cytometer (Partec, Germany) with a 488 nm laser for excitation of BODIPY and a 527 nm band pass filter for emission detection. Data analysis was performed using FCS Express V4 software (DeNovo Software, USA) selecting the cells with a FSC/SSC plot, thereby excluding cell debris. The gated events were analyzed by the fluorescent signal (FL1) expressed as median fluorescence intensity (MFI).
In order to proof intracellular localization of nanoparticles, confocal laser scanning microscopy (cLSM) experiments were performed on a LSM SP5 STED Leica Laser Scanning Confocal Microscope (Leica, Germany), consisting of an inverse fluorescence microscope DMI 6000 CS equipped with a multi-laser combination using a HCX PL APO CS 63 × 1.4 oil objective. Bodipy-labelled nanoparticles were excited with an argon laser (20 mW; λ = 514 nm), detected at 530–550 nm (pseudocolored green) and the cell membrane was stained with CellMaskOrange (2.5 μg mL−1, Invitrogen) using a laser DPSS 561 nm (≈1.3 mW), detected at 570–600 nm (pseudocolored red).
(i) Plain PSNPs (referred as PS): 6.0529 g of styrene, 250.72 mg of hexadecane, 103.55 mg of the initiator V-59 and 5.99 mg Bodipy-1 were added to 24 g of water containing 0.6 g of Lutensol AT50.
(ii) Carboxy-functionalized PSNPs (referred as PS-COOH):: 5.88286 g of styrene, 0.15258 g of AA, 251.28 mg of hexadecane, 99.71 mg of the initiator V-59 and 6.10 mg Bodipy-1 were added to 24 g of water containing 0.6 g of Lutensol AT50.
(iii) Amino-functionalized PSNPs (referred as PS-NH2):: 5.8887 g of styrene, 252.30 mg of hexadecane, 100.62 mg of the initiator V59 and 6.06 mg Bodipy-1 were mixed with 24 g water containing 0.12 g of AEMH and 0.6 g Lutensol AT-50.
After 1 h of stirring for pre-emulsification, the miniemulsion was prepared by ultrasonicating the mixture for 2 min by 450 W at 90% intensity (Branson sonifier W450 Digital, ½′′ tip) at 0 °C. For polymerization, the temperature was increased to 72 °C and reaction proceeded overnight.
In order to explore the effect of surface chemistry, we presented a qualitative and quantitative analysis of the protein corona of three differently surface functionalized PS-NPs and their influence on the composition of the protein corona. Furthermore, we proved that the uptake of un-functionalized PS-NPs pre-incubated with either soft corona or Apo-A1 was prevented. Understanding how NPs interact with proteins as well as the effect of the different corona morphologies and compositions on cellular mechanisms (like uptake and trafficking) are important factors in the design of NPs. This is particularly so for those NPs with regulated biological identities and physiological effects and also for experiments based on “personalized protein corona” intended for clinical applications as suggested by Hajipour et al.49
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c7nr02977b |
‡ Both authors contributed equally. |
This journal is © The Royal Society of Chemistry 2017 |