Article pubs.acs.org/accounts
In Situ Characterization of Protein Adsorption onto Nanoparticles by Fluorescence Correlation Spectroscopy Li Shang*,†,‡ and G. Ulrich Nienhaus*,†,§,∥,⊥ †
Institute of Applied Physics, Karlsruhe Institute of Technology (KIT), 76131 Karlsruhe, Germany Center for Nano Energy Materials, School of Materials Science and Engineering, Northwestern Polytechnical University, Xi’an 710072, China § Institute of Nanotechnology, Karlsruhe Institute of Technology (KIT), 76344 Eggenstein-Leopoldshafen, Germany ∥ Institute of Toxicology and Genetics, Karlsruhe Institute of Technology (KIT), 76344 Eggenstein-Leopoldshafen, Germany ⊥ Department of Physics, University of Illinois at Urbana−Champaign, Urbana, Illinois 61801, United States ‡
CONSPECTUS: Nanotechnology holds great promise for applications in many fields including biology and medicine. Unfortunately, the processes occurring at the interface between nanomaterials and living systems are exceedingly complex and not yet well understood, which has significantly hampered the realization of many nanobiotechnology applications. Whenever nanoparticles (NPs) are incorporated by a living organism, a protein adsorption layer, also known as the “protein corona”, forms on the NP surface. Accordingly, living organisms interact with protein-coated rather than bare NPs, and their biological responses depend on the nature of the protein corona. In recent years, a wide variety of biophysical techniques have been employed to elucidate mechanistic aspects of NP−protein interactions. In most studies, NPs are immersed in protein or biofluid (e.g., blood serum) solutions and then separated from the liquid for analysis. Because this approach may modify the composition and structure of the protein corona, our group has pioneered the use of fluorescence correlation spectroscopy (FCS) as an in situ technique, capable of examining NP−protein interactions while the NPs are suspended in biological fluids. FCS allows us to measure, with subnanometer precision and as a function of protein concentration, the increase in hydrodynamic radius of the NPs due to protein adsorption. This Account aims at reviewing recent progress in the exploration of NP−protein interactions by using FCS. In vitro FCS studies of the adsorption of important serum proteins onto water-solubilized luminescent NPs always showed a stepwise increase of the NP radius upon protein binding in the form of a binding isotherm, regardless of the type of NP and its specific surface functionalization. This observation indicates formation of a protein monolayer on the NP. Structure-based calculations of protein surface potentials revealed that positively charged patches on the proteins interact electrostatically with negatively charged NP surfaces, and the observed protein layer thickness always matched the known molecular dimensions of the proteins binding in certain orientations. Temperature and NP surface functionalization have also been identified as important parameters controlling protein corona formation. Notably, while the corona formed from a single type of serum protein was reversible, protein adsorption from complex biological media such as blood serum was entirely irreversible. These quantitative in vitro studies are of great relevance to the bio−nano community and especially to researchers developing engineered nanomaterials for biological and biomedical applications. Future efforts will be directed toward elucidating kinetic aspects of protein corona formation and the detailed structure of the adsorbed proteins at the molecular level. To better appreciate the biological responses triggered by NP exposure, more efforts will be devoted to the exploration of the biomolecular corona as it forms on NPs in contact with living cells, tissues, and even entire model organisms. These studies are challenging when performed in a well-controlled and quantitative fashion and rely on the availability of sophisticated analytical tools, particularly, quantitative optical imaging techniques including FCS and related fluctuation methods.
1. INTRODUCTION
the design and production of engineered nanoparticles (NPs) with well-controlled physicochemical properties. They can be tailor-made for specific medical and technological applications, and some of them have already found widespread use in
Although nanotechnology as a research field only dates back to the late 20th century, new nanotechnology concepts and applications are emerging at an impressive pace across a broad range of fields, including electronics, energy, and medicine, and are likely to have an enormous impact on human life and entire economies.1−5 Recent advances of nanotechnology have led to © 2017 American Chemical Society
Received: November 16, 2016 Published: February 1, 2017 387
DOI: 10.1021/acs.accounts.6b00579 Acc. Chem. Res. 2017, 50, 387−395
Article
Accounts of Chemical Research consumer products.6,7 Their concomitant release into the environment is unavoidable, leading to unintended human exposure to NPs. Serious risks to human health may result upon NP uptake through the lung, gut, or skin.8−10 Important examples of intended NP exposure in biomedical applications (mainly through intravenous injection) are contrast agents, for example, iron oxide NPs in magnetic resonance imaging, or drug carriers, such as lipid NPs in drug delivery.11,12 Upon incorporation by a living organism, NPs come in contact with biological fluids such as blood or lung epithelial lining fluid. Proteins and other biomolecules in the fluids rapidly bind to the surface of NPs, forming a biomolecular adsorption layer, commonly referred to as the protein corona (Figure 1).13−15 Although surface modification such as
increase in hydrodynamic radius of the NPs upon corona formation. In this Account, we present an overview of recent progress in exploring the protein corona around NPs by using FCS, highlighting several important findings based on this technique.
2. FCS MEASUREMENT OF PROTEIN CORONA FORMATION FCS measures the average duration of brief bursts of photons from individual fluorescence emitters diffusing through an observation volume of about 1 fL in a confocal microscope (Figure 2A,B).27,28 In the simplest approach, the temporal
Figure 1. Schematic illustration of protein corona formation around NPs and subsequent in situ versus ex situ characterization. Two different proteins are depicted in cyan and green.
PEGylation may reduce nonspecific protein binding, it is exceedingly challenging to completely suppress it.16,17 The properties of such protein-coated NPs differ significantly from those of the bare NPs, and their interactions with the biological system are mediated by the dense adsorption layer.18 Hence, further biological responses of the living system after NP uptake, including cellular internalization, biodistribution, and toxicity are largely defined by the nature of the protein corona.19−23 To anticipate such responses, we require a comprehensive understanding of the protein corona, including especially the following aspects: (1) kinetics and dynamics of corona formation; (2) identification of the types of adsorbed proteins and their abundance; (3) structural properties and biological effects of these proteins; (4) essential physicochemical NP properties governing protein corona formation (NP size, shape, surface chemistry, etc.); (5) relationship between the nature of the protein corona and the biological responses (in vitro, that is, in cultured cells, and in vivo). This knowledge is a prerequisite to properly assess nanotoxicity and to develop safe nanomedicine tools. The exploration of NP−protein interactions has attracted many scientists all over the world, as reflected by the near-exponential increase of the number of relevant publications over the past decade. A wide variety of analytical techniques have been developed to characterize protein adsorption onto NPs. By using ex situ methods (such as transmission electron microscopy, gel electrophoresis, and mass spectrometry), the corona can be studied after separating the NP−protein complexes from unbound proteins, whereas in situ methods (such as fluorescence spectroscopy, circular dichroism, and dynamic light scattering) monitor protein adsorption while the NPs are immersed in the solution (Figure 1).24−26 Fluorescence correlation spectroscopy (FCS) is one such in situ method that allows us to measure, with subnanometer precision, the
Figure 2. (A) Schematic depiction of a fluorescent molecule diffusing through the observation volume of a confocal microscope during an FCS measurement. (B) Example of a typical fluorescence intensity time trace. (C) Normalized FCS autocorrelation decay curves, shifting to the right with increasing size of the NPs due to protein adsorption.
autocorrelation function, G(τ), is calculated from the measured fluorescence intensity time trace, I(t), G (τ ) =
⟨I(t + τ )I(t )⟩ −1 ⟨I(t )⟩2
(1)
with lag time τ and the angular brackets denoting the time average. Given that the observation volume is well approximated by a three-dimensional (3D) Gaussian, the autocorrelation function for free 3D diffusion is given by −1/2 −1⎛ ⎛ r0 ⎞2 τ ⎞ τ ⎞ ⎜ 1 ⎛ ⎟ G (τ ) = ⎜1 + ⎟ 1+⎜ ⎟ ⟨N ⟩ ⎝ τD ⎠ ⎜⎝ ⎝ z 0 ⎠ τD ⎟⎠
(2)
Here, ⟨N⟩ is the average number of fluorescent entities in the observation volume; r0 and z0 denote the radial and axial extensions of the observation volume, which can be measured in a calibration experiment. By fitting the experimental G(τ) in eq 1 with this model function, we determine the diffusional correlation time, τD, which in turn yields the translational diffusion coefficient via D = r02/(4τD). Finally, the hydrodynamic radius, RH, is calculated by using the Stokes−Einstein relation, 388
DOI: 10.1021/acs.accounts.6b00579 Acc. Chem. Res. 2017, 50, 387−395
Article
Accounts of Chemical Research
Figure 3. (A) Structure of the HSA molecule and the HSA corona forming on the NP surface. (B) The hydrodynamic radius is plotted as a function of HSA concentration. The blue solid line represents a fit of the Hill binding model (eq 4) to the data. Adopted from ref 32.
RH =
kBT 6πηD
saturating conditions. Figure 3B shows that RH increases in a stepwise fashion with HSA concentration, indicating a limited loading capacity of the NPs. These data can be modeled quantitatively, assuming that HSA forms a monolayer around the NPs with a thickness ΔRH of 3.3 nm. Structurally, HSA resembles an equilateral triangular prism, with sides of ∼8 nm and a height of ∼3 nm. Thus, the ΔRH of 3.3 nm implies that the proteins adsorb via their triangular surfaces onto the NPs. Once a monolayer of ∼20 HSA molecules is formed, the NP size remains constant up to the solubility limit of HSA; there is no tendency to further accrete protein. Remarkably, HSA monolayer formation has been observed in all our FCS experiments, without a single exception, on NPs with different cores (FePt,32 CdSe/ZnS,33 Au34) and different surface functionalities (carboxylic acid, amine,35 lipid,36 PEG polymer,16 zwitterionic ligand37). Other important plasma proteins of different size and charge such as transferrin (Tf),38 apolipoprotein A-I,31 apolipoprotein E4,31 and C3 protein36 also exhibited monolayer formation on NP surfaces. As for HSA, the monolayer thickness was always found to coincide with molecular dimensions of the adsorbed protein, as inferred from X-ray structure analysis or other biophysical techniques such as analytical ultracentrifugation, viscometric and fluorescence studies.31 These well-controlled FCS studies of the adsorption of one type of protein to NPs are very revealing but do not emulate the real biological situation of protein adsorption from complex biological fluids such as human serum. Therefore, we have recently extended our FCS measurements to study protein adsorption onto CdSe/ZnS QDs immersed in human blood serum.39 Despite the presence of ∼3700 different proteins in serum,40 we still obtained well-defined binding curves as in Figure 3B, indicating formation of a thin protein layer of welldefined thickness on QD surfaces, independent of the type of QD surface ligand.39 We note that FCS was also used by Rädler and co-workers41 to study Tf adsorption onto polystyrene NPs with RH > 50 nm. These experiments, however, differ from the ones described here in that the proteins rather than the NPs were fluorescently labeled. The amounts of free and NP-bound Tf were determined by decomposing the autocorrelation function into contributions from protein and NP diffusion, and deviations from a simple binding model were interpreted in terms of additional, weakly bound adsorption layers. While we cannot completely exclude effects due to different NP sizes,42,43 we reiterate that we have never found evidence of multiple protein adsorption layers in our direct FCS determinations of RH via diffusion of fluorescently labeled NPs. Serum proteins have been designed by evolution to be colloidally very stable. Therefore, as long as the protein structures are not disrupted in
(3)
with Boltzmann constant kB, temperature T, and viscosity η. Consequently, the increasing NP size due to protein adsorption results in an increase of τD, reflected by a shift of the autocorrelation curve toward longer times (Figure 2C). In all experiments performed as yet, we have observed a stepwise response of the NP radius change as a function of protein concentration, which can be described by using a Hill binding model, ⎛ ⎜ cNmax R H = R 0 ⎜1 + KD ′ ⎜ 1 + [P] ⎝
⎞1/3 ⎟ n⎟ ⎟ ⎠
( )
(4)
where R0 is the radius of the bare NP, c = VP/V0 is the volume ratio of protein and NP, Nmax is the maximum number of adsorbing proteins, n is the Hill coefficient controlling the steepness of the curve, [P] is the concentration of free protein in solution, and KD′ is the midpoint concentration of the binding curve (KD′ = n KD for a binding reaction of n ligands and one NP, with equilibrium dissociation coefficient KD). For simplicity, the FCS experiments are preferentially carried out with a large excess of proteins over NPs, so that the concentration of free protein in the solution is negligibly decreased if proteins adsorb onto the NPs. FCS experiments yield structural information on the protein adsorption layer with subnanometer precision. High-quality FCS data require careful instrument calibration and data acquisition procedures, as previously discussed.29 Of note, excellent colloidal stability of the NPs and NP−protein conjugates is also a key prerequisite. The overall radius increase measured by FCS provides important clues about the protein orientation on the NP surfaces, whereas KD′ reveals the binding affinity of proteins to NPs. We mention in passing that more elaborate versions of fluctuation spectroscopy have been utilized for protein corona studies, such as dual-focus FCS (2fFCS), a variant of the conventional FCS method that provides an absolute calibration standard of the confocal volume.29−31
3. FORMATION OF A PROTEIN MONOLAYER AROUND NANOPARTICLES In 2009, we presented the first FCS analysis of human serum albumin (HSA, Figure 3A) adsorption onto small (∼10 nm diameter) polymer-coated, fluorescently labeled FePt NPs and CdSe/ZnS core−shell quantum dots (QDs).32 Most striking was the observation of a single protein monolayer under 389
DOI: 10.1021/acs.accounts.6b00579 Acc. Chem. Res. 2017, 50, 387−395
Article
Accounts of Chemical Research a major way due to adsorption, the exposed surface of the first protein layer should have near-native properties and, specifically, a weak aggregation tendency.
The intriguing revelation of a close structure−affinity relationship opens a promising new perspective for chemical approaches to modulate protein corona characteristics by altering the charge distribution on the protein surface. For example, we have shown that the HSA corona on NP surfaces can be significantly changed by modifying the surface charge characteristics of HSA via succinylation or amination.33,45 Whereas the protein corona of native HSA on negatively charged dihydrolipoic acid-protected QDs (DHLA-QDs) is 3.3 nm thick, binding of succinylated HSA to DHLA-QDs results in a significantly thicker HSA corona of 8.1 nm (Figure 4).
4. ELECTROSTATIC INTERACTIONS GOVERN PROTEIN ORIENTATION ON NP SURFACES Proteins are biopolymers with abundant charged groups on their surfaces, and NPs typically also carry charges on their surfaces to endow them with colloidal stability. Therefore, electrostatic forces are expected to crucially contribute to NP− protein interactions. Importantly, at typical ionic strengths of biological media (∼150 mM), the Debye screening length is
