The bio-corona and its impact on nanomaterial toxicity

Introduction

Throughout evolution, humans have been exposed to airborne nanosized particles, which are currently defined as materials with at least one external dimension in the size range of approximately 1–100 nm. However, such exposure has increased dramatically over the last century due to anthropogenic sources and is expected to further increase due to the rapidly developing field of nanotechnology. The applications of engineered NM are not only increasingly used in technical products but also more and more in biotechnology and biomedicine (1–3). Engineered nanoparticles (NPs) can be generated by a variety of physical, chemical, and/or biotechnological methods, and additionally tailored by specific functionalization procedures to optimally meet their intended technological or biomedical applications (3, 4). As such, NPs have a wide range of potential applications in biomedical sciences to include targeted drug delivery systems and contrast agents for imaging methods such as magnetic resonance imaging or fluorescence spectroscopy and even as theranostics (5–16).

Though, besides the current enthusiasm for nanotechnology, the wide use of nanomaterials may pose unknown risks to human health and the environment (7, 17–21). Particularly, experiences with ‘new’ materials in the past, such as asbestos, have sensitized the public also for the risk of ‘similar’ NM, such as single-walled carbon nanotubes (SWCNT). Issues relevant to pulmonary toxicity of SWCNT, their direct cytotoxic effects, their ability to cause inflammation, and induce oxidative stress upon inhalation are currently actively investigated (7, 17–21). As NPs are able to highjack the endocytosis machinery and may enter almost any cell type (22), their recognition and engulfment by macrophages, phagocytosis and bio-distribution in tissues and the circulation are still under intense discussion (7, 17–21). Once inside a cell, NPs may cause adverse effects (23) resulting even in permanent cell damage (24, 25). As potential mechanisms, oxidative stress, inflammation, genetic instability, and the inhibition of proper cell division have been suggested (26–29). Moreover, immunosuppressive effects of NM and their impact on increased sensitivity of exposed individuals to additional microbial infections as well as their effect on the microbiome are currently investigated.

The discussion about nanosafety upon intended as well as unintended exposure of humans and the environment is thus clearly important and still ongoing (7, 17–21). As society at large is now aware of the use of NM in ever growing quantities in consumer products and their presence in the environment, critical interest in the impact of this emerging technology has grown. The main concern is whether the unknown risks of engineered NP, in particular their impact on health and environment, outweighs their established benefits for society. Therefore, a key issue in this field is to evaluate their potential toxicity, identify responsible pathways of toxicity (PoTs), and nano-structure activity relationships (nanoSARs). As the biomolecule corona forms rapidly in exposure relevant complex physiological and natural environments, an understanding of its relevance for nanotoxicity is mandatory.

Protein corona takes shape

When NPs are designed for specific biomedical applications, such as drug delivery and imaging purposes, these often requires intravenous injections of the NPs (2, 30). It is not surprising that for this intended human exposure, the detailed knowledge about physical and chemical aspects associated with the behavior of NPs in physiological systems in general has been recognized as an important factor especially for understanding nano-toxicology (19, 31). In physiological environments NPs are exposed not only to relatively high ion concentrations or drastic pH changes (32), but also to a huge variety of complex biomolecules (33). Many body fluids (e.g. blood, lung lining fluid, saliva, intestinal juice, etc.) have one common characteristics, their composition is highly complex (34, 35). Thus, whereas the physico-chemical properties and behavior of NPs can be engineered and controlled in technically stable, protected environments, such as technical products, this is no longer the case in complex physiological or natural environments. Such ‘complex environments’ are not only represented by simple and higher organisms, including humans, but also by complex solid and liquid interfaces to which NPs are exposed during their technical application and/or their intended/unintended exposure of humans. Moreover, complex organisms are again composed of various additional ‘complex (micro)environments’, such as organs and cells, which also differ very dramatically in their physico-chemical composition.

In this complex (micro)environments, NPs adsorb various (bio)molecules due to their high surface energy (36–38). The physical and chemical interactions with proteins and/or other biomolecules (e.g. phospholipids, sugars, nucleic acids, etc.) will in most cases significantly affect the NPs’ behavior and fate. By examining databases, it is obvious that the ‘nanoparticle corona’ represents a still unresolved ‘hot topic’, with high scientific and economic relevance. Over the last 10 years the number of publications dealing with ‘nanoparticles’ and ‘corona’ increased almost 15-fold (Figure 1). The adsorbed proteins clearly defines NPs surface and mediates further interactions between the NPs and the biological environment (39–42). This coating marks the biological identity of NPs and indirectly causes its ‘transformation’ by drastically altering the NPs’ colloidal stability. Here, the protein corona can either have a stabilizing effect by inducing steric stabilization (43) or a de-stabilization impact, caused by protein mediated bridging, charge compensation and/or by introduction of charge inhomogeneity onto the NPs surface. Upon aggregation, multiple interactions may result in stronger affinities compared to proteins binding to single NPs, which is more likely to occur in a biological solution in which particles are highly diluted. Moreover, there could even be a trapping of (abundant) proteins in such aggregates with low or no affinity for single NPs. Depending on the NPs, such aggregation may also require a certain time and thus, additional may impact the kinetics of corona formation. Ex vivo, the sub-fractionation of this aggregates by centrifugation techniques are possible during NP synthesis, but these effects are almost impossible to control in vivo, not to mention to predict. In summary, the aggregation of NPs will add an additional level of complexity, which has to be considered in the description and application of NPs within physiological systems (36, 37, 44).

Figure 1:

Timeline of PubMed entries matching the search criteria ‘nanoparticles’ AND ‘corona’ from 2005 to 2014.

The term ‘protein corona’ was introduced to the nanoparticle community by the study of Cedervall and colleagues (40), and stimulated a whole field of investigations. Until today, the ‘hard corona’ represents an analytically accessible protein/biomolecule signature of a NP in a certain environment, and this term is often used to describe the long-lived equilibrium state, representing a protein signature of a NP in a certain environment (36, 37, 45, 46). On top of this ‘hard corona’ some models suggest the existence of a ‘soft corona’, a putative more loosely associated and rapidly exchanging layer of biomolecules (37, 40, 46–48). However, since this ‘soft corona’ desorbs during current purification processes, its existence and (patho)biological relevance remains vage. Hence, as inspection of the current literature revealed that the term ‘soft’ versus ‘hard’ corona seems to mostly create more confusion than helping to describe and resolve scientific questions, we suggest to generally refer to analytically accessible NP-protein complexes simply as the ‘protein corona’.

Factors influencing NM-corona composition, -fate, and -toxicity

The biomolecular corona primarily interacts with biological systems and thereby constitutes a major element of the NP’ biological identity affecting multiple molecular-scale interactions (36, 37, 46, 49, 50). A significant difference in the bio-physical properties between such corona-covered NPs and those of the formulated pristine particle during manufacturing is observed (17, 36, 37, 46, 51, 52). Ccientist and eventually regulators therefore consider (bio)molecule-coated NMs as novel materials with different properties compared to the pristine nanomaterials during production (17, 36). The NP-protein complexes significantly influence the the particles biodistribution in patients as well as the subsequent biological responses of the body, potentially contributing not only to favorable nanomedical reactions but also to unwanted (patho)biological side-effects (17, 36, 53–55). Clearly, for the rational development of NM for any kind of biological or biomedical application but also for the unwanted uptake of NPs it is thus the key to understand the formation and kinetic evolution of the protein corona (37, 38, 56–59). Numerous studies have been conducted to generally dissect and mechanistically understand the biomolecule corona on nanoscale materials, its dependence on the nanoparticles’ physico-chemical properties and its biomedical and/or (patho)biological relevance (17, 36, 44, 53–55). Typically, corona profiles differ significantly from the protein composition of the (biological) fluid investigated (36, 43, 44, 46, 60, 61). Distinct proteins will be either enriched or display only weak affinity for the nanoparticle surface. The determination of the corona by the protein source is also an important factor for the so called ‘personalized protein corona’ (PPC) (62). Therefore, humans with a specific disease may have specific NP coronas, which could be a determinant factor in nano-biomedical science (62–64). As we are still at the beginning of understanding the role of the protein corona for a biomedical application, the issue ‘PPC’ adds an additional level of complexity to the field, which certainly has to be addressed and most importantly confirmed in future studies.

However, the relation between original surface functionality of the NPs and the nature of the corona is far from being trivial and currently still remains impossible to predict in complex physiological environments (36, 43, 44, 46, 55, 60, 61). The different physic-chemical properties, such as material size and surface properties, but also the relative ratio of the physiological fluid to the nanoparticle dispersion and the exposure time play an important role for the composition and evolution of the protein corona, although the underlying physical mechanism are not yet resolved in detail (Table 1, and references therein) (36, 37, 43, 44, 46, 55, 60, 61, 84). Moreover, when NPs move from one biological (micro)environment to another, e.g. from the blood system via different cellular uptake mechanisms into cells (e.g. monocytes or macrophages), a key issue is whether the original corona remains stable or is subjected to substantial changes (44), which adds an additional level of complexity. Untill today it is assumed that after passing through several ‘biological (micro)environments’, the final corona may still contains a fingerprint of its history and keeps a memory of its prior journey through the body (44, 70).

Though, some studies still tend to suggest of having identified ‘the major NM factor’ controlling protein/biomolecule corona formation. However, as convincingly shown by recent comprehensive studies, none of the above-mentioned factors, such as the NPs’ physicochemical properties or exposure time, alone is able to determine formation and composition of the protein/biomolecule corona (36, 46). In the study by Tenzer and colleagues (36), bioinformatic unsupervised hierarchical cluster analysis revealed that protein corona profiles were correctly grouped according to exposure time as well as the NPs’ physicochemical properties. The relation between the pristine NPs characteristics and the nature of the corona in complex environments is thus far from being trivial (36, 43, 44, 46, 55, 60, 61). Despite the complexity and analytical challenges of the biomolecule corona during its ex situ characterization, researchers are facing additional challenges during its in situ analysis in both, physiological and environmental systems.

However, as recent data demonstrate that the plasma protein corona is surprisingly stable and matures only quantitatively rather than qualitatively (36), one might however hypothesize that the corona may not be subjected to significant changes, even when passing through several ‘biological (micro)environments’, unless processing is performed by enzymatic cellular machineries. Particularly for metal and metal oxide NPs, dissolution processes have been recognized of being essential for the NPs’ fate, biodistribution and also ROS-mediated toxicity (46, 85–89). Here, proteins play central roles not only by forming a protein corona, but also by binding released toxic metal ions, as has been shown for silver NPs in both, physiological and environmental systems (86, 90). Close inspection of the literature indeed reveals that the detailed fate of the original corona, as it travels through membranes and barriers, thereby interacting with the extracellular matrix and various cellular enzymatic machineries is still not resolved in detail, and may differ in various body organs, such as the liver or the brain. By developing nanomaterials for in vitro diagnostic sensors or drug delivery-/cell-targeting vehicles, one has to at least consider the formation and potential impact of the biomolecule corona on the biomedical performance of the product.

As described above, the protein corona is surprisingly stable, however, it is important to clarify the presentation of functional biomolecular motifs of the individual proteins in the corona, as the nanoparticle–biomolecule complex interacts with cells and biological barriers, and thus, engaging with different biological pathways. This question so far, has only been addressed in single-protein models, but needs to be clarified in the future. Kelly et al. have recently presented an approach to map protein binding sites on the biomolecular corona which will help to understand the spatial location of the proteins and their binding sites after binding to NPs (91).

Summarizing our current knowledge, a multi-parameter classifier will be required to generally model and predict nanoparticle-protein/biomolecule interaction profiles in complex physiological and environmental systems in the future.

The blood system – Model for corona formation and toxicological impact

Aiming at therapeutic drug delivery and/or imaging systems, NPs are expected to be most commonly intravenously injected into the blood stream, at which point they immediately begin to interact with a very complex biological milieu. Also, if NPs are capable of overcomming biological barriers, such as the lung, the skin or the gut, they will likewise face the bloodsystems prior to their potential transport to distict organs. Here, a variety of biomolecules such as lipids, sugars and especially proteins will adsorb onto the surface of NPs. Such coronas may change how the body interacts with a NP because the nanoparticle’s size and surface characteristics such as charge or targeting molecules can be altered (36, 44, 55, 60, 92). Indeed, particularly the plasma protein corona has been shown to be highly complex (36). Proteins involved in physiological as well as toxicological relevant processes in the blood system, such as complement activation and coagulation have been identified in the coronas of various NPs (Figure 2) (36, 44, 55, 60, 66, 92, 93). Identified proteins span about three to four orders of magnitude dynamic range, most likely covering most of biologically relevant corona proteins (36). Notably, the respective abundance of all of these proteins was affected by the NPs’ characteristics, such as size and surface functionalization, and plasma exposure time. As NPs’ uptake is also an important determinant for nanopathology and targeted delivery (25, 44, 94), several reports showed that the protein corona has a major impact on the NPs’ cellular uptake. Particularly, the recent study by Walkey and colleagues indicates that indeed distinct protein corona signatures are able to predict cellular uptake of gold and silver NPs (46). Hence, covering NPs with a ‘physiological coating’ can indeed promote or inhibit their interaction with the cellular uptake machinery, whereas the surface charge of the bare NPs appears to be less important (36, 94–96).

Figure 2:

NM- and exposure time-dependent adsorption of pathophysilogically relevant human plasma proteins. Profiles of proteins involved in complement activation and coagulation are shown. Exposure time-dependent functional classification of corona proteins on negatively charged silica NPs (SNP). SNP of various sizes and various surface-functionalization were analyzed. Abbreviations: CO8B, Complement component C8 beta chain; CO6, Complement factor C6; CO7, Complement factor C7; CO9, Complement factor C9; CO8A, Complement component C8 alpha chain; C4BPA, C4b binding protein alpha chain; C1S, Complement C1s subcomponent; CFAB, Complement factor B; CO4B, Complement factor C4B; C1R, Complement C1r subcomponent; CO5, Complement C5; CO4A, Complement factor C4A; CFAH, Complement factor H; CO3, Complement C3; HABP2, Hyaluronan binding protein 2; ANT3, Antithrombin III; FIBA, Fibrinogen alpha chain; KLKB1, Kininogen 1; MMRN1, Multimerin 1; FA12, Coagulation factor XII; IC1, Plasma protease C1 inhibitor; HRG, Histidine rich glycoprotein; PLMN, Plasminogen; KNG1, Kininogen 1; THRB, Prothrombin; VWF, von Willebrand factor; FA5, Coagulation factor V; TSP1, Thrombospondin 1.

Consequently, one may use this information to rationally engineer the uptake-properties of NPs by modulating corona fingerprints. Different apolipoproteins have been described to promote transport across the blood-brain barrier (97) and different immunoglobulins and complement factors, known as opsonins, enable uptake into monocytes (30) while dysopsonins like albumin and again apolipoproteins (40) inhibit uptake. However, these findings are mostly based on the prior knowledge of selected proteins based on their biological function in isolation. As also functionalization of NPs with proteins seems not to (completely) prevent corona formation, the complexity of the protein corona with more than hundred different proteins, makes it difficult to predict the impact of individual proteins in vivo (36, 44, 46, 98). Hence, the engineering of modified coronas by depletion or enrichment of protein groups is required as the next step to identify corona components causally involved in (cell type specific) NPs’ uptake. Cleary, obtaining comprehensive quantitative and qualitative protein corona signatures, and ideally the implementation of an international standardized corona profile database resource, will finally allow the bioinformatic analysis and exploitation of signatures to guide a subsequent rational in vitro/in vivo investigation of the potential (adverse) impact of corona proteins in physiological systems (36, 44, 46, 60).

Albeit several studies reported impacts of corona formation on NPs’ exposure of the blood system, most of these effects were described to occur at rather late exposure time points (66, 99, 100). Employing primary human cell models of the blood system, it was though shown that already the early corona formation affected toxicological processes at the nano-bio interface (36). The study showed that although the studied pristine NPs existed only for a short period in the blood system, these were still able to affect vitality of endothelial cells, trigger thrombocyte activation and induced hemolysis (36).

Hence, it was concluded that formation of the biomolecule corona rapidly modulated the NPs’ decoration with bioactive proteins, thereby protecting cellular components of the blood system against NP-induced (patho)biological processes, and in addition also influenced cellular uptake of NPs (36). However, whether these findings are valid for other NP formulations as well as for NM of other shapes and materials remains to be investigated.

Prevention strategies to inhibit corona formation

The protein and most likely also other biomolecular coronas are currently a still unpredictable complex factor, potentially triggering not only desired reactions but also undesired toxicological biological responses (36, 44, 46, 55). Hence, there are currently numerous attempts to chemically (completely) prevent protein adsorption, which also have been reviewed previously (16, 44, 61, 74, 101–103). In this models the NPs are functionalized with certain polymer chains, such as the addition of various polyethylene glycol-based chains (‘PEGylation’) onto the NPs’ surface, which are often referred to be highly ‘biocompatible’, as unspecific interactions with biological components are minimized (74, 101, 102). Under physiological conditions, NPs functionalized with PEG confers colloidal stability caused by interparticular repulsion (74, 101, 102). However, even complex ‘PEGylation’ is unable to completely prevent protein/biomolecule corona formation, albeit the extent of protein adsorption is clearly reduced (16, 44, 61, 74, 101–103). As protein adsorption is reduced, it is assumed that numerous cellular responses are affected, including opsonization by cells of the reticuloendothelial system (RES) (3, 44, 61). Thus, the circulation time in the blood system as well as the biodistribution of NPs may be modulated via PEGylation, although the detailed mechanisms are not yet resolved (66, 99, 100).

An alternative method to the PEGylation is the functionalization of the NP surface with zwitterions (104). As recently discussed, NPs with tunable hydrophobicity were designed which do neither adsorb proteins at moderate levels of serum proteins nor do they form hard coronas at physiological serum concentration. This innovative strategy may lead to new options for analyzing the interaction of NM with biosystems without any interference from protein binding. However, further development of the fabrication of ‘corona-free’ NPs is sufficient (104).

Again, standardized corona profiling combined with data mining and subsequent experimental verification is required to answer these questions. Also, extensive surface functionalization with biomolecules, such as antibodies tailored to achieve specific targeting of cell types and/or organs, or ‘cellular uptake proteins’ does not completely prevent corona formation, albeit such modifications clearly affect corona profiles. In some cases, protein corona formation was even considered a major factor significantly reduced cell-targeting efficacy in vitro as well as in vivo (97, 105–107).

Collectively, although sophisticated surface modifications reduce the adsorption of biomolecules to NPs, association with biomolecules does still occur. To our knowledge, there is no existing nanomaterial functionalization strategy, which will completely prevent the formation of a biomolecule corona in complex environments. The design and synthesis of such nanodevices represents certainly one of the key challenges required not only to finally understand the regulation and impact of the biomolecule corona but also to allow novel nanomaterial applications.

Corona complexity and evolution

Indeed, not only physico-chemical properties of NPs, but also the exposure time of NPs to biological environments as well as additional factors has been reported to affect the protein and most likely other biomolecule coronas. Such in situ transformation of NM may lead to altered biodistribution and directly or in directly impacts the efficacy of desired therapeutic and (patho)physiological responses (36, 38, 44, 105, 108) (Table 1). A current model suggests that particularly at short exposure times the ‘soft’ protein corona is initially formed around the NP, which is highly dynamic and subsequently rather slowly matures to the ‘hard’ corona by significantly changing its composition over time (36, 44, 72). Most studies however focused on corona formation upon prolonged nanomaterial exposure to complex biological environments (36, 44, 46, 72). These studies often failed to recognize that physiological systems are highly dynamic and might need to react instantly to external stimuli. In the blood system, flow velocities are heterogenous, such as high in the ascending aorta or slow within tumors. Also, processes controlling hemostasis and thrombosis need to be triggered within minutes or even seconds (66, 99, 109, 110).

The formation and evolution of protein layers on flat surfaces was first analyzed by Vroman (111), describing a time-dependent composition of the bio-coating, in which highly abundant proteins adsorbing only weakly dominate the early state. These adsorbed proteins are later replaced by less abundant proteins, which however bind with higher affinity, resulting in adsorption and displacement step (75, 111, 112). However, one has to keep in mind that the ‘Vroman-effect’ was demonstrated only for a mixture of a few proteins and is unable to predict protein binding kinetics to NPs in complex mixtures (111, 112). Nevertheless, several models used the ‘Vroman-effect’ to directly explain the evolution of the protein corona around NPs even in complex environments, resulting in the concept of a ‘dynamic protein corona evolution’ (44, 61, 70, 72). However, in complex physiological liquids, such as blood, containing more than two thousand different proteins, a high-resolution time-resolved knowledge of NP-specific protein/biomolecules adsorption is required, as various protein/biomolecules are expected to display increased or reduced binding over time. Indeed, snapshot kinetic proteomic profiling recently demonstrated the existence of complex protein adsorption kinetics (36). As predicted from the ‘Vroman-effect’, protein groups displaying increased or reduced binding over time were observed also in the complexity of human plasma (Figure 3) (36). Interestingly, novel protein binding kinetics, such as ‘cup-‘ or ‘peak’ shaped binding kinetics were observed (Figure 3), which cannot be solely explained by the ‘Vroman-effect’ (36).

Figure 3:

Correlation analysis to demonstrate distinct kinetic protein-binding modalities during the temporal evolution of the plasma protein corona. Time-smoothed normalized protein abundance profiles of negatively charged polystyrene NPs. NPs’ coronae were classified into four groups by correlation analysis and relative values were normalized to the maximum amount (set to 1) across all time points for each protein. Protein groups “rise” and “fall” showed increasing or decreasing binding over time, respectively. “Peak2” proteins, display low abundance at the beginning of plasma exposure and at later time points, but higher (peak) abundance at intermediate time points. “Cup” proteins show the opposite behavior, with a high abundance at early and late time points, but low abundance at intermediate time points. A selection of representatives is displayed. t: Plasma exposure time.

A potential reason why such complex binding kinetics have been unnoticed so far is the fact that most kinetic studies did not employ sensitive quantitative LC-MS-based proteomics. Thus, a protein, which was no ‘longer detectable’ at a certain time point, was classified as being ‘absent’ or ‘disappearing’ in previous studies, thereby contributing to the model of a highly dynamic protein corona (44, 61, 70, 72). Hence, we again want to emphasize that it is highly important to use the highest technological standards and standardized experimental procedures (SOPs) for the determination of protein corona binding kinetics (82), allowing inter-laboratory comparison and model building by further systematic studies. The same study also demonstrated that the plasma protein corona is highly complex, containing over 200 different proteins, and surprisingly established in less than one minute (36). In contrast, previous studies suggested that the protein corona has a rather low complexity, consisting of only a few tens of proteins, even when NPs are introduced into highly complex environments, such as the human blood, and evolved rather slowly (33, 44, 61, 72). Moreover, the study also showed that the corona composition changed almost exclusively quantitatively but not qualitatively over time (36). Previous models however proposed a highly dynamic protein corona, changing significantly in its composition over time due to continuous protein association and dissociation events (44, 61, 72).

Impact of NM-coronas on nano-toxicity

Nanotoxicology was proposed as a new branch of toxicology to address the gaps in knowledge and to specifically address potential adverse health effects caused by NM (19). Due to their small size, NM are characterized by a high surface area to volume ratio, rendering them highly reactive. The latter potentially results in yet unknown toxicity mechanims due to novel interactions of NM with biological systems, including the environment (19). In biological environments, NPs bound with proteins can result in physiological and pathological changes, but the mechanisms and potential NM-specific stress and toxicity pathways remain to be fully elucidated. Several studies in the past have addressed the possible impact of the corona on subsequent cellular responses. Liu et al. could show that in the absence of proteins the level of p-p38 was significantly elevated by the positively charged mesoporous silica NPs (MSNs), whereas negatively charged MSNs resulted in marked reactive oxygen species (ROS) production (113). Surprisingly their study showed that the presence of protein efficiently mitigated the potential nano-hazard (113). Shannahan et al. showed that AgNPs with or without a protein corona were able to induce a concentration-dependent cytoxicity and that all corona-coated AgNPs were found to activate cells by inducing IL-6 mRNA expression (114). Ge et al. demonstrated that the competitive binding of blood proteins on single-wall carbon nanotubes (SWCNT) influenced cellular pathways and resulted in reduced cytotoxicity that depended on the presence of protein adsorption and the highly competitive binding of blood proteins on the SWCNT surface (115). Their study suggested that the binding of various protein types onto the SWCNT surface can elicit different cytotoxic cellular response (115). Hu et al. reported a protein corona-mediated reduction of cytotoxicity of graphene oxide (GO). Here the interaction with different concentrations of serum proteins resulted in a decrease of cytotoxicity, whereas the direct interaction with bare GO nanosheets led to a physical damage of the cell membrane (116).

Although protein corona formation can mitigate nanotoxicity, the immune cell response, e.g. the complement systems, can be activated, depending on the type of corona formation, which may cause inflammation and damage to the host (117). For such a scenario, different groups demonstrated that by modifying the surface functionality and availability of reactive functional groups of NPs, the complement activation was attenuated (118, 119). Although NP-protein coronas in most cases appear to reduce cytotoxicity, immunotoxicity can be mitigated or activated depending on the type of NM and/or adsorbed biomolecules.

Systematic strategies to dissect the impact of NM-coronas on nano-toxicity

Determination of qualitative and quantitative (kinetic) corona profiles is a prime and critical step in order to understand the biological and toxicological impact of the nanoparticle corona on living systems. However, these data have to be complemented by biological assay systems of varying complexity. Albeit in traditional toxicology a ‘top down’ approach is often prefered, we feel that a rational hypothesis and knowledge-driven ‘bottom up’ strategy is needed to mechanistically understand (toxicologically) relevant processes at the nano-bio interface. Hence, we here introduce a suitable tiered systems biology-driven approach.

TIER 1: The routes by which NPs may enter the human body, and potentially elicit (adverse) effects, are understood to include inhalation, injection, ingestion and permeation through (diseased) skin. Conventional cell culture models representing these targets were developed almost a century ago and have demonstrated a significant value in biomedical research and safety testing of chemicals. Thus, simple or highly advanced in vitro models mimicking major exposure and application routes of NPs in mammals need to be used for identifying and dissecting basic structure activity relationships (SARs).

Currently, cell models are exposed to pristine or corona-covered NPs often in low throughput applications. Assays are performed either by pre-coating the NPs with biomolecules or by performing the experiments in biomolecule containing liquids, such as plasma- or FCS-containing cell culture medium. Albeit such studies provided basic insights into corona-mediated (toxic) effects, their low throughput and lack of standardization make inter-laboratory comparisons often difficult, sometimes even leading to contradictory results (3, 44, 46).

To overcome these problems, high-throughput screening (HTS)/high-content screening (HCS) experimental systems should be used. Cell based HCS has evolved dramatically, allowing HTS applications to measure the responses of cells to chemicals and, as recently described, also to NPs (36, 85, 120–122). The concept and technological execution of such assay systems have been reviewed (120, 121, 123), albeit the impact of the biomolecule corona was not specifically addressed so far.

The workflows used predominantly for such HCS/HTS screening approaches mostly focused on classical plate reader-based assays using biochemical readouts (120, 121, 123). However, the ability to automate the capture and analysis of fluorescent images of thousands of cells has made fluorescence microscopy an additional tool of systematic cell biology, applicable also for investigating nanobiology. Importantly, HTS assays can be automatically performed in microtiter plates, and thus, the analysis is highly economical regarding cells and NPs, and can be executed under stable SOPs.

As an example for such a HTS workflow to investigate the impact of the plasma protein corona on cell-death, the results are shown in Figure 4. By using a HTS-fluorescence microscopy imaging platform (108), we established a dual-color fluorescence cell vitality assay. By employing fluorescent probes that recognize cell viability by measuring intracellular esterase activity (calcein-AM; green) as well as plasma membrane integrity (ethidium homodimer-1/EthD-1; red), the assay allows for the simultaneous quantitation of live and dead cells.

Figure 4:

HTS-quantification of the silica NP-corona’s on toxicity. Cell viability was determined by the ratio of the average calcein (living cells; green) to ethidium homodimer-1 (dead cells; red) signals. Only living cells are able to convert the virtually non-fluorescent cell-permeable calcein-AM to the intensely fluorescent calcein, resulting in an intense uniform green fluorescence of living cells. EthD-1 is however excluded by the intact plasma membrane of living cells, and only enters cells with damaged membranes. Here, it undergoes a 40 fold enhancement of fluorescence upon binding to nucleic acids, thereby producing a bright red fluorescence characteristic for dead cells. Colon epithelial CaCo-2 cells were exposed to AmorSil30 NPs suspended in DMEM (open rectangles) or DMEM containing 10% human plasma (filled rectangles) for 4 h and analyzed by automated microscopy using the Cellomics ArrayScan^® VTI. A minimum of 500 cells were analyzed per well, and each treatment was done in triplicate. Whereas the ratio of living to dead cells remained almost unchanged for the untreated Ctrl. as well as after the treatment with 0.6 μg/mL or 6 μg/mL AmorSil30 in the presence and absence of proteins, incubation with 60 μg/mL or 600 μg/mL in absence of proteins led to a significant decrease of this ratio, indicative of cell death. Columns, mean; bars, ± SD from three independent experiments.

Combining systematic analytical and cell-based technologies with the advances in bioinformatics is accepted as a powerful approach to rationally dissect and understand cause-effect relationships of NPs in living systems. Thus, analytical and experimental data delivered by proteomic and in vitro HTS profiling experiments, can be collected in a data repository. This data repository can be used for the identification of corona-dependent structure activity relationships (CoroNanoSARs) linking protein coronas with NPs’ characteristics and biological and toxicological effects. Importantly, recent developments intend to combine and to mine data obtained by methodological divergent ‘omics-technologies’, such as proteomics, transcriptomics, metabolomics, cellomics and epigenomics, already allow theoretically to build a multi classifier score for each type of NP under investigation (124, 125). Moreover, biological pathway exploitation will provide rational information, which type of NP affects corona-dependent cellular responses for the subsequent identification of key corona proteins causally involved in the observed (adverse) biological effects.

Nevertheless, the design and execution of such HTS assays is demanding, concerning technology platforms, experimental workflow and experience in data acquisition and handling.

TIER 2: Conventional 2D-cell monocultures are ideal for HCS/HTS approaches and already provided first important information regarding NanoSARs and/or pathways of toxicity (PoTs) (36, 85, 108, Nel, 2012 #4332, 121, 122). Nevertheless, the relevance of such results needs to be subsequently confirmed by using cell models mimicking more closely the in vivo exposure situation. Limitations of cell monocultures lay in the differentiated functions of many cell types in communication with their natural neighboring (micro)environment and thus, in the accurate prediction of in vivo tissue function. Shortcomings of 2D setup are the lack of mimicking the complex three-dimensional (3D) (micro)environment, in which the cells and an extracellular matrix exist in an organized structure. In this regard, efforts were shifted toward developing multiple 3D culture systems that can better recapitulate in vivo tissue functions. Compared with 2D cell cultures, 3D models are expected to achieve better capturing of signaling pathways and response to NPs (126). Hence, in vitro co-culture systems of higher complexity are valuable tools to verify the relevance of mechanistic insights obtained from more simple experimental systems and to further evaluate the effects of NPs on organs and human health, allowing cell-to-cell communications, inter-/intra-signaling regulations, as well as inter-/intra-NPs trafficking absent in monocultures (127, 128). Albeit such systems have not been applied so far to analyze the impact of the biomolecule corona on PoTs, these systems are in principle suitable, albeit only with low-throughput. In a transwell insert, different cell types, like endothelial, epithelial and immune cells can be cultivated in the same well allowing cellular communication via soluble second messengers (127, 128). The communication between endothelial cells and cells that have the direct contact with pristine or corona-covered NPs may play an important role in the systemic effects of NPs and thus, may result in a more realistic judgement regarding the impact of the protein corona. In addition, recent technological developments allow to develop ‘organ-on-a-chip’ systems to move further towards mimicking complete tissues (129). Albeit being far from trivial, such models have a high potential to unravel involved toxicity mechanisms. Also, they will serve as a valuable tool for the rational planning of subsequent in vivo studies, and also allow to avoid unnecessary animal experiments as requested by the 3R rule.

TIER 3: Similar to assessing the toxicity of ‘simple’ chemicals, animal experiments will ultimately be needed to verify the in vivo relevance of corona-nano-structure-activity mechanisms and also to predicte their relevance for human exposure and/or biomedical applications (17, 19, 31, 130). To date, various animal models are in principle available to investigate and dissect the relevance of the NPs’ biomolecule corona for human health. These are ranging from rodents to fish and other organisms, and have been reviewed before (17, 19, 31, 131–135). Notably, recent developments also applied successfully HTS/HCS approaches for whole animals, such as zebrafish (121, 136). However, comprehensive reports on studies focusing particularly on the biomolecule corona are missing so far. Clearly, the researchers are facing various experimental and analytical challenges, when investigating not only ‘simple’ endpoints such as vitality, but when analyzing the corona-mediated fate and biotransformation of NPs in vivo. Here, the recovery of NPs from various organs under ‘corona-preserving experimental conditions’ will require the development of novel extraction protocols and imaging techniques.

Conclusion and outlook

The biomolecule corona is far from being an already resolved topic in basic as well as in applied nanoscience, including nanotoxicology. Even regulatory agencies begin to accept that corona signatures might be relevant for (improved) NM’ risk assessment and prediction. Although the relationship between nanomaterial design and physiological responses has been studied intensely for almost two decades, only some general principles have emerged. Fortunately, the biomolecule corona is now no longer ignored as an ‘unknown factor’, but acknowledged as a potential but yet fully unexploited opportunity to understand and predict the impact of NPs on physiological environments. Also, the corona complexity increases once we are considering additional ‘small inhabitants’ of the human body, i.d., the human microbiome. Such interactions between NM, bacteria, and cells as well as the role of biomolecule coronas have not been addressed so far.

Currently, we cannot yet predict how the synthetic identity of a nanomaterial influences the structure, composition and evolution of the protein corona. However, presenting a cell or an organ with a NP, we do understand that it does not see the bare NP, but the particle with an entire surrounding biomolecule corona profile. Still unresolved are the questions which corona proteins are involved in the cell recognition, uptake and toxicity, and which mechanism governing the interaction of the NM, pristine or corona-covered, with cells. Furthermore, it has to be resolved whether every corona protein or only a certain subset is accessible to and/or biologically active in the physiological and toxicologically relevant environment. Without such a detailed knowledge, a rational design of NMs interacting only with defined corona proteins and cells in a controlled and predictable way remains challenging.

It is now accepted that biomolecule coronas are established rapidly, and formation most likely occurs to varying degrees for all nano-sized materials in general. Hence, the (long term) existence of pristine NPs in complex physiological environments appears to be rather an exception to the rule. Corona profiles are highly complex, consisting of a variety of adsorbed proteins potentially capable of modulating biological responses, and are ex situ rather stable. Nevertheless, long-term effects and the relevance of coronas on the fate and transformation of NPs deposited in vivo, such as organs, require future attention. Quantitative, high-resolution LC-MS/MS is now capable to dramatically reduce biomolecule corona characterization time and can provide quantitative data from large libraries of NM. In order to comprehensively analyze corona profiles and to mechanistically understand the coronas’ biological/toxicological impact, a tiered multidisciplinary approach is clearly mandatory/requisite, including not only comprehensive analytical methods but also the involvement of high throughput/high content ‘omics’ technologies together with bioinformatic data mining to reach the next level. Despite the achievements of the past years, the establishment of nanostructure activity-relationships linking NP/corona properties to physiological or toxicological responses remains still a distant goal. However, such knowledge is ultimately needed not only to understand and minimize nanotoxicity but also to develop NM allowing an improved and safe application of nanotechnology in general.