Particle toxicology and health - where are we?

Particle Toxicology

Particle (and fibre) toxicology focusses on understanding and describing the relationship between an exposure to the agents and their capability to adversely affect human health and aims to identify the underlying pathobiological mechanisms. There is a long history of particle toxicology that was recently summarized by Donaldson and Seaton [2]. This substantial body of information on quartz, coal and asbestos has been useful to identify metrics that can be used to predict adverse health effects for new challenges presented, as for example by engineered nanomaterials with structures in the nanoscale (between 1 to 100 nm) and especially nanofibers and nanoparticles (NPs) with two or three dimensions, respectively, in the nanoscale [3–5]. The famous fibre paradigm originates from our understanding on how asbestos can cause lung cancer including mesothelioma. It includes issues like frustrated phagocytosis where the macrophages are not able to fully engulf the fibres, resulting in sustained generation of reactive oxygen species, as well as blocking of the stoma in the chest wall by fibres with subsequent inflammatory responses [6]. The many cases of silica exposure have exemplified that this substance can cause irreversible lung disease known as silicosis. Unlike asbestos, where much effort is put into preventing exposures, silica exposure continues to occur in the 21st century which goes hand in hand with the occurrence of silicosis. The importance of particle solubility was illustrated by silicic acid released from the quartz surface as being the component responsible for the toxicity. At the same time we know that the physical aspects of particles and fibres will have a major role in mechanical irritation in the lungs resulting in inflammation. Although inflammation in itself can be seen as a host response that should protect us, excessive inflammation, especially if it persists over prolonged time, can cause fibrosis, oedema as well as the formation and progression of tumours.

Following the cases of asbestos, coal and silica/quartz, people became very much aware of the health risks associated with being exposed at work to particles and fibres, including man-made mineral and synthetic vitreous fibres. By the end of the previous century (i.e. the late 1990s), scientific attention shifted from occupational exposure to environmental exposures as a number of epidemiological studies indicated that ambient PM exposures were associated with premature mortality and worsening of several diseases. Apart from the known lung diseases like asthma and chronic-obstructive pulmonary disease, cardiovascular diseases and cancer were noted to be impacted by exposure to outdoor PM. The levels of PM at which the associations were reported were well below prevailing air quality standards, and particle toxicology was needed to prove causality and provide plausible biological mechanisms that could explain and support these associations [7]. Although 2-3 decades of toxicological research have taught us a lot, there is still a lot of uncertainty on how low levels of particles can cause so many different effects in humans, and a ranking of substances or sources of emissions can be presented to guide risk management and policy measures is still missing [8].

There are some common features to particle-induced hazard. For example, inhalation of particles can cause oxidative stress which may lead to genotoxicity and reversible or persistent inflammation [9]. Some of the target organs, such as the lungs and the cardiovascular system, are well studied. More recently, it became evident that inhaled particles can also affect the central nervous system [10, 11] and reproduction [12]. This seems to be by and large due to translocation of the smallest particles in the mixture, referred to as ultrafine PM or nanoparticles, into the internal organs [13]. The evidence derived from studies using ultrafine PM has also led to a rapidly growing interest in the toxicology of man-made, manufactured or engineered NPs in the last 15 years. Interestingly, there is a significant overlap in the toxicology of ultrafine particles and engineered NPs and cross talk between these two areas can boost our understanding of the toxicology of nano-sized particles [14].

Particle Medicine

The field of particle medicine started to emerge about 50 years ago [15, 16]. Research intensified when it became clear that nanoparticles (NPs) have unusual properties with exciting potential for the improvement of established and the development of novel clinical applications. Areas of high interest are imaging and diagnostics, drug delivery and anticancer therapy [17–25]. To date, more than a dozen anticancer nanomedicines have been approved for clinical use, and almost 40 are in clinical trials. Particular attention is currently paid to precision medicine. Here, in case of cancer treatments, the goal is to develop personalized anticancer nanomedicines, which are engineered to target, for instance, a particular type of tumour with a specific location and microvasculature pattern.

Anticancer nanomedicines have various beneficial properties. These include enhanced accumulation in solid tumours with reduced off-target delivery and improved safety. The enhanced permeability and retention (EPR) effect is thought to play a central role in the improved efficacy. The EPR effect is based on leaky blood vessels and impaired lymphatics in tumour tissue, which leads to enhanced extravasation of NPs into tumour tissue and reduced clearance by lymphatic drainage [26–28]. However, the EPR effect would only be relevant with respect to relatively large and well-vascularized tumours, and various challenges remain, including the treatment of leukaemia and metastatic disease. Such challenges must be addressed by more specific targeting strategies. These include surface modifications of NPs with cancer cell targeting ligands, such as folate or antibodies, and magnetic targeting [26, 29–31]. Specific targeting of nanomedicines may also be helpful for addressing the challenge that the EPR was effective in animal models, but it failed, so far, to perform well in human patients [32, 33].

Surface modifications are not only important for targeting, but also for preventing clearance of nanodrugs by phagocytotic cells from the reticuloendothelial system. This is critical for increasing the circulatory half-life of nanomedicines, and for preventing damage of non-target tissues due to activation of resident phagocytotic cells. The classical strategy for reducing interactions with the reticuloendothelial system is to modify the surface of the medical NPs with polyethylene glycol (PEGylation), which hinders opsonisation (binding of proteins recognised by phagocytotic cells), but may shield targeting ligands [34]. Other strategies include biomimetic coating with proteo-lipid membranes extracted from leukocytes [35], platelets [36] or other cells, and autologous cells should be used for clinical applications. Such membranes contain biological functions based on their protein content, which include “marker of self/don’t eat me” signals. A more defined approach is to modify the NP surface with markers of “self/don’t eat me” signals such as the anti-phagocytotic SIRPα ligand CD47 and respective synthetic peptides [37].

Tuning the shape and size of NP medicines is also critical. These features influence the radial drift (margination) of NPs in blood vessels. Whereas small spherical NPs accumulate within the center of blood vessels, disc-like particles display enhanced lateral drift due to tumbling, and also have larger surface areas for endothelial adhesion [38]. Both are important for extravasation in tumour tissue. In addition, NP density appears to influence margination, with enhanced margination of high-density NPs. This leads to an easier distribution of low-density NPs in the body, which is associated with more rapid renal clearance [39].

Renal clearance is also strongly dependent on NP size, shape and charge, not only due to indirect effects based on margination, but also due to the properties of the renal filtration barrier. Rigid spherical NPs are not efficiently cleared by renal filtration if their hydrodynamic diameter exceeds 5.5 nm [40]. Surprisingly, large nanofiber-like materials, such as individualized carbon nanotubes with diameters and lengths of up to 20 – 30 nm and 500 – 2000 nm, respectively, easily cross through the renal filtration barrier when aligned in the right orientation, and are cleared with similar efficiency as small molecules [41, 42]. Tuning of renal clearance of nanomedicines must be carefully adjusted to keep the balance between maintaining therapeutic plasma levels and safe elimination from the body.

During the last decades, nanomedicines with a wide range of structural features have been developed for the delivery of small molecule drugs, biologics, nucleic acids, or co-delivery of multiple compounds [24, 43]. The first approved nanomedicines for drug delivery were a synthetic polymer conjugated to the anticancer protein neocarcinostatin [44], and PEGylated liposomal doxorubicin (Doxil/Caelyx) [45]. Since the 2000s, polymeric micelle-based nanomedicines for drug delivery have been approved for clinical applications. Polymeric micelles consist of amphiphilic block copolymers that self-assemble into a core-shell structure. The hydrophobic core can be loaded with hydrophobic small molecule drugs or biologics, whereas the hydrophilic shell can be further modified by PEGylation or with targeting ligands.

Until now, polymeric micelles or other types of NPs, which did not have any anticancer activity themselves, were used as vehicles for the delivery of anticancer agents. An exciting more recent development are micellar nanocomplexes (MNCs) for anticancer protein drug delivery, which have anticancer activity themselves (also in the absence of the anticancer protein) [46]. The MNCs were based on derivatives of the green tea catechin (-)-epigallocatechin-3-O-gallate, which has known anticancer activity. After loading of the MNCs with the anticancer protein drug Herceptin, synergistic anticancer activity of the MNCs and Herceptin has been observed, which resulted in enhanced antitumor activity in vitro and in vivo compared to Herceptin alone [46]. In addition, prolonged plasma half-life and enhanced accumulation in the tumour tissue of the MNCs compared to the protein drug alone were demonstrated [46].

The most important goal for the future is to achieve personalized treatments by designing precision nanomedicines. Crucial for achieving this goal is an improved understanding of the interactions between NPs and biological structures and tissues. This will help to guide the development of smart strategies for manipulating NP surface chemistries, which is required for reducing unspecific interactions and for increasing targeting of tumors with specific, individual properties. Replacement of conventional synthetic polymers by bio-inspired molecules, such as antibodies for targeting and “don’t eat me” signals, will play a central role in developing smarter and personlized strategies. Important will be also the tuning of the size and shape of nanomedicines based on the understanding of how these features influence interactions in normal and disease states., Smarter strategies for tuning size and shape will include further development of multistage and tumor environment-responsive NPs, with larger parental NPs that have improved circulatory half-life, which release smaller NPs at the tumor that can easier penetrate into the tumor tissue [47, 48]. Addressing the variability between patients as a function of age, gender, ethnicity, their individual disease state and other patient-specific factors will be essential to guide the design of NPs for precision nanomedicine and to achieve personalized treatments.

Dynamic Fate of Inhaled Nanoparticles in the Lungs of Rodents

Deposition of inhaled NPs is governed by their diffusivity in the air leading to a rather homogeneous deposition density on the epithelium of the various regions of the respiratory tract. As a result, the deposited fraction in the lungs reaches a maximum of about 50% of the inhaled aerosol at a size of 20 nm [49, 50] with about 15% alveolar deposition and 35% depositing in the conducting airways of the head and thorax. In other words, two thirds of the deposited 20 nm sized NP will be cleared rapidly within 24 hours by mucociliary action within the ciliated conducting airways and one third will be long-term retained in the lung periphery. Below that size, increasing fractions deposit in the airways of the head and thorax according to their increasing diffusivity with decreasing size, such that less NP reach the distal alveolar region (see Fig. 1). In this short summary a few consequences on the biokinetics fate will be discussed for insoluble NP.

Relocation and re-entrainment of Inhaled NPs in the rodent lungs

In rodents, there is evidence that rather constantly about 80% of micron-sized particles can be lavaged from the lung surface during six months post-exposure (p.e.) when the lavaged particle fractions are normalized to the contemporary lung burden at each time point [51–53]. In contrast, only 20% of 20-nm-sized NPs and about 30% of 70-nm-sized NP deposited during inhalation or instillation can be lavaged – see Fig. 2 [54–59].

The low NP recovery becomes plausible since NPs deposit rather uniform on the surface of an alveolus by diffusion. Indeed, macrophages on the epithelial surface will rapidly phagocytize all NP which happen to be deposited close to where a macrophage happens to reside, while distant NP are not recognized due to the weak opsonizing signal of NPs. However, since only about 10 – 100 macrophages are residing in an alveolus of 300 nm diameter of a healthy rodent [60], these surface macrophages cover only a fraction of <0.001 of the alveolar surface [57]. Since epithelial Type I+II cells possess well-identified endocytotic mechanisms of NP uptake and less so for micron-sized particles their endocytic NP uptake capacity competes with the phagocytic capacity of surface macrophages. As discussed in our previous paper [59], numerous authors provided evidence that epithelial type 1 cells (EC1) can endocytose NP at the epithelial surface and eventually exocytose those towards the baso-lateral side at the end of the last century, including the review of Lehnert [61] and papers by Adamson and Bowden [62–64]. More recently, Thorley and co-workers [65] provided evidence for prominent NP uptake by EC1 while they ruled out uptake by alveolar epithelial type 2 cells (EC2) and the passage of NPs by paracellular transport was also discounted. They stated that NP uptake occurs preferentially by diffusive processes into the cytoplasm which allows for exocytosis and transport across the basal membrane into the interstitium, as supported by in vivo studies. In collaborative studies [66, 67], localization of AuNP and TiO₂-NP in interstitial spaces was demonstrated by Transmission Electron Microscopy (TEM). According to [68, 69] the cytoplasmic leaflets of EC1 provide the largest portion of their surface area on the basal membrane which separates the adjacent vascular endothelial cells to allow unhindered gas exchange. Yet, data up to 24 hours do not support the notion that neither AuNP nor TiO₂-NP cross EC 1 at both - the “active site of gas exchange” [68] and the basal membrane, since any NP exocytosis would lead to rather rapid uptake by vascular endothelial cells and translocation into the circulation which was not observed. Instead the translocated AuNP or TiO₂-NP fractions to blood were rather small during the first 24 h p.e.. Hence, EC1 either exocytose directly into septal interstitial spaces – which provide only a relatively small surface area at the side of EC1 – and/or the exocytosed NP migrate in between the basal membrane and the EC1membranes to the next septal interstitial space. Once there, NP may be phagocytized by interstitial macrophages (IM), fibroblasts, etc. Referring to [61, 70] there is a large population of IM such that the role of IM in phagocytizing NPs and long-term retention in the septal interstitial spaces appears plausible. For gradual re-entrainment back onto the epithelium, Lehnert [61] reviewed the pathway of IM onto the epithelium. In addition, IM and fibroblasts may exchange their NP-load with AM like in a relay. Hence, NP behave differently compared to micron-sized particles which are retained on the surface of the rodent epithelium to be eliminated by long-term, macrophage-mediated clearance (LT-MC) at a rate of initially 2-3%/d which declines gradually over time [54, 71, 72]. Surprisingly, NPs are cleared with a similar clearance rate dynamics via this macrophage-mediated transport indicating re-entrainment of the NPs from the interstitial spaces back on top of the epithelium for subsequent LT-MC towards the larynx and into the gastro-intestinal-tract (GIT) [55, 56]. Re-entrainment may occur across the alveolar epithelium and/or via interstitial-lymphatic clearance to bronchus-associated lymphoid tissue (BALT) entering the epithelial surface at the airway epithelium for clearance by surface macrophages, as hypothesized several decades ago [52, 73–76]. NP passage through BALT at bronchioles-alveolar duct junctions back onto the bronchiolar epithelium [63] [77] cannot be excluded; however, BALT may play an important immunogenic role for fluid absorbed from the alveolar surface, but the reverse flow onto the epithelial surface was postulated in the literature but has not been proven so far [68, 69]. Furthermore, there are only about 30-50 BALT sites in the rat lungs [61, 78] which are far too few compared to the number required for NP re-entrainment and LT-MC clearance.

As a result, LT-MC is the most prominent long-term clearance mechanism for insoluble NPs from the peripheral lung of rodents. In fact, although the NP are retained in the septal interstitial spaces close to blood vessels, only rather small fractions are translocated via this pathway into circulation leading to subsequent accumulation in secondary organs and tissues which, however, depends strongly on the physicochemical properties of the NPs. For example, four different materials (iridium (Ir), elemental carbon (EC), TiO₂, and gold (Au)), which were inhaled as freshly generated 20-nm NP aerosols during a single 1–2 h exposure by healthy, adult, female Wistar–Kyoto rats; the translocation percentages (normalized to the alveolar NP deposition) of IrNP (7.96 ± 0.47) and TiO₂NP (6.95 ± 0.14) were significantly higher than those of elemental carbon (2.18 ± 0.31) and AuNP (1.79 ± 0.39) as shown in Table 12.1 of [79].

Retention in secondary organs was followed up to six months after the inhalation of the IrNP aerosol showing no detectable clearance [55, 58]. During chronic exposure to insoluble NP, continuous accumulation is likely to occur in secondary organs. This may play a modulating role in adverse cardio-vascular health effects which have been observed in epidemiologic studies after exposure to ambient fine and ultrafine particles [80].

NP pathways from lungs to circulation and accumulation in secondary organs and tissues

Systemically circulating NPs may accumulate in secondary organs and tissues by two particle clearance pathways, (i) NP translocation across the air-blood-barrier (ABB) either directly into blood circulation or via the thoracic lymph duct and (ii) NP absorption across the GIT walls, again either directly into blood circulation or via the thoracic lymph duct, of those NP which were eliminated from the lungs towards the larynx and swallowed into the GIT. The latter clearance pathway has a fast component of those NP deposited on the epithelium of the conducting airways which are cleared rapidly by mucociliary action (MCC) followed by a slow component of those NP from the peripheral lungs eliminated by LT-MC towards the larynx.

The contribution of both pathways towards secondary organs was quantitatively investigated for the first time in a series of studies in which identical 70-nm-sized TiO₂NP were applied to rats either as a single bolus via intratracheal (IT) instillation or via gavage (oral ingestion) or via intravenous injection. Their biokinetics were determined quantitatively in the entire organism during the next 28 days [81–83]. The biokinetics data obtained from the gavage study were used to estimate the absorbed TiO₂NP fractions across the gut walls after IT-instillation which had been cleared from the lungs via the larynx into the GIT. In Fig. 3 the ratios Ri of gut-absorbed and subsequently accumulated TiO₂NP divided by the sum of both – gut-absorbed and ABB-translocated TiO₂NP - are shown for liver, spleen, kidneys and the carcass (comprising of skeleton, muscles, fat, skin, but without organs) and the integral absorbed TiO₂NP fraction at different time points between 1 and 28 days after IT-instillation. The integral absorbed TiO₂NP ratios increase with time up to 0.2 of all systemically circulating TiO₂NP due to the continuous LT-MC transport leading to continuous absorption across the gut walls. Ratios in liver, kidneys and the various tissues of the carcass stay below 0.05, but the absorbed TiO₂NP ratios in the spleen are about 10.1 at days 1 and 7. These data show that accumulation in secondary organs and tissues is predominantly determined by ABB-translocated TiO₂NP which, however, occurs mainly during the first few days after IT-instillation (see Fig. 3). Yet, with increasing retention time the gut-absorbed NP fractions become more and more important.

Interrelated Concepts of Exposure, Dosimetry and Dose-Metrics for NP Risk Assessment

Results of numerous studies, in vitro and in vivo, have revealed that engineered NPs and ambient ultrafine particles (UFPs, e.g. diesel exhaust) can induce significant dose-dependent toxicity in primary and secondary organs. In order to characterize appropriately the associated risk as a function of hazard and exposure, exposure-dose-response relationships have to be established. With respect to inhalation as the main route of exposure - involving acute, subchronic or chronic rodent studies - a minimum of three exposure concentrations plus sham-exposed controls should be used [84]. The experimental design should include detailed aerosol characterization and biokinetic data. Essential for the evaluation of the results of inhalation studies is to determine the retained dose (lung burden) at the end of exposure to establish dose-response relationships. This is a fundamental requirement which is often forgotten when reporting results only as exposure-response data. Expressing the dose by different metrics such as particle mass, surface area, volume, or number, will provide additional information about potential underlying toxicological mechanisms that control outcomes. The revised OECD Test Guideline 413 [85] for subchronic rodent inhalation now includes a requirement to determine retained lung burdens.

Establishing toxicologically well-characterized particles as positive and negative “Benchmark Materials” [86] for a comparative Hazard and Risk Characterization will facilitate the grouping of inhaled particulate materials which are tested in subacute, subchronic or chronic inhalation studies. This involves comparing the slope of dose-response curves (potency) or the No Observed Adverse Effect Levels (NOAELs) of rodent studies to obtain a hazard ranking. If only a Lowest Observed Adverse Effect Level (LOAEL) can be identified in the rodent study, benchmark dose (BMD) analyses are appropriate. Expressing the retained lung burden by different dose-metrics (particle mass; surface area; number) will help to identify the most appropriate metric by comparison to toxicologically well-characterized benchmark materials. For example, if the measured response to different particle sizes of the benchmark material shows the same correlation with a chosen dose metric, then this metric has better predictive value than other dose metrics. The more predictive metric, then, should be used for evaluating the unknown nanomaterial in comparison to the benchmark material. For poorly soluble particles, surface area has been found to be of best predictive value [87, 88].

As a first step of risk assessment using results of a rodent study, the exposure–dose–response relationship can be analyzed by using a BMD approach in order to derive an associated benchmark concentration (BMC) as a “safe” exposure level for rodents [89]. Results derived from a rodent study can then be the basis for risk extrapolation to human exposure scenarios by deriving a Human Equivalent Concentration (HEC), provided that species differences in respiratory tract dosimetry and retention kinetics are considered [90]. The HEC is defined as the exposure concentration predicted by modelling to result in the same normalized retained lung burden as measured in rodents after acute, subchronic or chronic inhalation. Normalization of deposited or retained dose is frequently done using species-specific lung weight or alveolar surface area. Effects, however, may be different between the species due to different sensitivities or mechanisms of uptake and effect. To account for this and possible toxicokinetic/toxicodynamic differences, assessment or uncertainty factors may have to be included [91].

Dosimetric extrapolation of results from rodent inhalation studies for deriving the HEC is achieved with the MPPD (Multiple Path Particle Dosimetry) Model [92]. Important recent refinements of MPPD include improved input values for allometrically adjusted respiratory parameters [93] and the choice of specific rat and mouse strains. A suggestion of how to extrapolate a NOAEL and associated exposure concentration (NOAEC) determined in a subchronic rodent inhalation study to a chronic 2-year study has been proposed by [94, 95]. The approach is shown in Fig. 4, it involves the use of the MPPD model to estimate a NOAEC which, after two years of exposure, results in the same NOAEL that was determined in the subchronic study. Such dosimetric extrapolation will avoid the use of an uncertainty factor for extrapolation from subchronic exposure to chronic exposure.

The deposition of airborne particles is affected by the effective or actual density of aerosols, which makes it an important input variable for the MPPD model. Specifically, if the aerosol consists of agglomerated and aggregated NPs, the void spaces between the individual particles of an aerosol cluster change the effective aerosol density to be significantly less than the material density. Whereas a number of methods have described how to measure effective aerosol density (e.g., [96–99], a simple approach for measuring the actual density present in a rodent inhalation study is to perform a short-term inhalation, sacrifice the animals at the end of exposure and measure the deposited lung burden. This allows calculation of the deposition fraction. One then runs the MPPD model with rat specific and body weight allometrically adjusted inputs and changes the input value for the density iteratively until it fits the calculated deposition fraction. Work on this approach is ongoing.

Among the intrinsic physico-chemical properties that impact on NP toxicity, surface properties are, in theory, most influential because of the direct interaction of the particle surface with cells and subcellular components. In addition, extrinsic or functional NP properties, such as specific surface reactivity and dissolution rate are important for categorization and grouping of NPs. With respect to dissolution, NPs are often grouped by their solubility in water [100]. However, water solubility is not always appropriate for predicting in vivo dissolution. Obviously, dissolution rates of NPs can vary widely, which has to be considered because the biopersistence will be affected depending on the dissolution rate. There are two basic approaches to determine solubility/dissolution of particles in cell-free systems: the static system, determining the equilibrium solubility, and the dynamic system, determining dissolution rates. The latter mimics more closely the in vivo situation, whereas the former can reflect in vitro conditions. The composition of the dissolution fluid is another critical factor with respect to closely simulating in vivo conditions, e.g., phagolysosomal fluid (pH ~4.5) or epithelial lung lining fluid (pH ~7.4). The significance of reliably assessing the in vivo dissolution rate of NPs lies in the possibility to characterize their biopersistence by estimating the overall pulmonary clearance rate and retention halftime. This will be necessary for subsequent use as inputs into the MPPD model for purposes of human risk extrapolation modelling.

Furthermore, indications from acellular assays that dissolution in the respiratory tract takes place raises the question about the fate and underlying mechanisms by which tissues respond to the dissolving NPs. Bioprocessing or biotransformation events investigated using High Resolution Transmission Electron Microscopy (HRTEM), Scanning transmission electron microscopy (STEM) and Electron energy-loss spectroscopy (EELS) analyses revealed subcellular NP dissolution and chemical interactions resulting in secondary very small NPs [101]. Amorphous nano-silica (SiO₂) in alveolar macrophages examined by HRTEM after subchronic inhalation in rats at 27 days post-exposure period had undergone significant in vivo breakdown and transformation [102]. In particular, a portion of the original NPs were partially dissolved and secondary SiO₂-reaction zones (precipitates) formed as a result of in vivo processing (also called bioprocessing – see section “The 5Bs” below) as shown in Fig. 5 ([102], previously not published). Compared to the starting materials, the bioprocessed SiO₂ particles showed dissolution patterns (voids/pore formation) and outward growth of reaction zones. The degree of in vivo processing of NP can be evaluated with HRTEM which, coupled with dose-response monitoring, could provide further information for NP risk assessment.

As a look to the future, exploration of the physicochemical changes at the particle surface over time following exposure, i.e., bioprocessing, will improve our understanding of tissue dosimetry, biodistribution, and, ultimately, the mechanisms by which inhaled particulates exert toxicological effects.

The 5 Bs: Bioavailability, biopersistence, bioprocessing, biomodification, and bioclearance of nanoparticles and the role of the biomolecule corona

For drugs and molecular chemicals, the proportion of a drug or other substance which enters the circulation when introduced into the body, and so is able to have an active effect, is considered the bioavailable dose, while the length of time that a molecule (typically a toxicant) remains in the biological organism without being metabolised or excreted is termed biopersistence. Clearly, the route of introduction will affect this proportion, with direct intravenous injection resulting in close to 100% bioavailability, while oral or nasal introduction for example, might result in lower bioavailability due to the biological barriers that exclude some of the compound. Given some of the unique aspects of NPs, however, such as their tendency to agglomerate at higher concentrations [103], their tendency to interact with biomolecular constituents of their surroundings to form so-called biomolecular coronas [104–106], and the size-modulated cellular uptake mechanisms [107, 108], the bioavailable dose for NPs is poorly understood compared to that for molecular substances, leading to enormous challenges in determining meaningful dose-response curves [109, 110]. Additionally, the dynamic nature of NPs and their tendency to evolve and transform during storage [111] or upon interaction with media [112] makes dose characterisation challenging. Indeed, due to interactions with their surroundings, the dispersion media play an important role in defining the NPs dose, but also in determining the NPs “biological” identity, which is the set of proteins and other biomolecules associated with the NP and specifically the outer surface of the absorbed corona which forms the interface to engage with biological receptors [113–116]. Indeed, proteins such as Bovine Serum Albumin (BSA) have long been used as agents for dispersal of NPs for toxicity assessment [117], as has its environmental equivalent, natural organic matter (NOM) for ecotoxicological studies [118, 119]. However, in some cases, adsorption of biomolecules induces partial agglomeration of the NPs thus altering the bioavailable dose [120]. The absorbed biomolecule corona can also alter the biopersistence of the NPs, for example, by changing the rate of dissolution of NPs. For example, the corona adsorbed to NPs has been shown to slow dissolution by blocking oxidative processes [121] or by promoting sulfidation process and entrapping nanocrystals of Ag₂S in the hard corona [122], while in other cases, adsorbed coronas have been found to accelerate dissolution, especially where there is a strong affinity for the metal ions by the proteins and an excess of the proteins [123].

Recent work using the model organism Daphnia magna has shown that proteins secreted by the Daphnia into the media induce some agglomeration of NPs but since this brings the NP-agglomerates into the size range of Daphnia’s normal food source, it appears to increase uptake / bioavailability of the NPs [108]. Similar findings have suggested that cellular response to the presence of NPs also involves secretion of proteins in response to the initial form of the NP taken up by the organism, causing the “initial” corona to evolve, which can lead to altered uptake and impacts not currently considered in assessing toxicity [124]. Thus, NPs interacting with living organisms are dynamic systems where the NPs affect the organism and the organism affects the NPs. This suggests that there is more than one biological identity for each NP and that this identity evolves as the NPs interact and are internalised and processed by organisms and cells. Indeed, early work to model the corona evolution experimentally, whereby NPs were sequentially incubated in biological fluids representing the external environment (serum) and the cellular environment (e.g. cytosolic fluid) indicated that the resulting corona had proteins from both fluids [125]. .

The physicochemical changes that particles undergo in tissues following exposure, which include dissolution, deagglomeration, secondary particle formation, etc., have been termed bioprocessing. As noted above, the highly reactive surface area of NPs leads to adsorption of a biomolecule corona, which plays a key role in the subsequent bioprocessing of NPs, determining for example the rate of dissolution, the uptake mechanism, and the subsequent biodistribution. From the discussion of the biomolecule corona above, it is clear that the decision as to which bioprocessing step occurs is determined by the biomolecule signals that are associated with the NPs. For example, it has long being recognised that certain proteins trigger recognition by phagocytic cells, leading to rapid clearance of NPs associated with these proteins, which are known as opsonins [126]. Examples of proteins known to enhance phagocytotic uptake include collectin molecules such as surfactant protein A (SP-A) and SP-D, as well as members of the complement cascade involved in wound healing. On the other hand, enormous effort in nanomedicine has been devoted to the development of so-called stealth NPs that can evade the immune system and avoid uptake by phagocytes, either by reducing overall protein binding or by selectively binding the so-called deopsonising proteins such as albumin [34]. Clearly, the specific proteins that bind to NPs will be influenced by the route of internalisation, with the lung surfactant proteins being the main candidates for binding following inhalation, serum proteins being the first binders in the case of intravenous exposure, and a range of enzymes and food biomolecules being potential corona constituents for gastric exposure [127]. Other transformations included under the broad term bioprocessisng include enzymatic digestion [128], such as has been reported for carbon nanotubes [129] and graphene materials [130].

Studies of the biodistribution of NPs as a function of exposure route correlate with this assumption that biomodification, as determined by the NP-associated biomolecules, strongly influences the distribution of the NPs. For example, in vivo studies using radiolabelled gold NPs in rats indicated that different exposure routes led to different biodistributions of the NPs, which is most likely a result of different biomolecule adsorption and thus different bioprocessing signals [131, 132]. Also, radiolabelled Au NPs in different sizes (1.4-200 nm) exposed by intra-oesophageal instillation into healthy adult female rats resulted in detectable NPs (ng/g organ) in the stomach, small intestine, liver, spleen, kidney, heart, lung, blood and brain after 24 h as measured by gamma-spectroscopy, with the highest accumulation in secondary organs being for the smallest particles, while the 18 nm particles showed a higher accumulation in brain and heart compared to other sized particles [132]. On the other hand, Au NPs delivered tracheally to rats resulted in the majority of NPs remaining in the lungs (> 95% of the initial dose, ID) with < 1% of the ID translocated to the kidneys, liver, blood and urine, and < 0.01 of the ID reaching the spleen, uterus and heart [131]. While these studies did not explicitly attempt a comparison on the basis of adsorbed biomolecules, it is clear that such a study, including recovery of the NPs and assessment of their biomolecule coronas following translocation and final localisation, would shed important new light on the biomodification of NPs and the role of the biomolecule corona in the bioprocessing steps. Indeed, the hope for both nanomedicine and nanosafety is to “design” NP surfaces to acquire the desired corona to direct the bioprocessing to minimise the risk of harm to humans. Designing the NP surface to tailor the corona is already underway, via the design of stealth particles as discussed above, or indeed via surface modification with small molecules that induce protein-misfolding in a component of the NP-associated protein corona, and which enhances or reduces the NPs’ susceptibility to cell-specific receptor-mediated endocytosis [133]. Such unfolding could lead to unintended immune responses though via display of so-called cryptic epitopes [134], and thus such approaches needed to be conducted with care.

The term biomodification is normally used to indicate some modification of a biological organism, e.g. via genetic or mechanical means, while we use the term to understand the impact of the NPs on the biological organism‘s biochemistry. Thus, we attempt to distinguish between physicochemical transformations of the NP that occur post uptake (biotransformation, bioprocessing) and cellular processes that result in the incorporation of the NP degradation products into existing biological pathways.[128–130] Biomodification pathways, via which the degradation products can be incorporated into existing biological pathways, are especially interesting in terms of the design of safer NPs. One example of a biomodification pathway, proposed for iron oxide NPs, showed that 10 nm iron oxide NPs were degraded in macrophages and the resulting free iron was transformed to ferritin and hemosiderin iron-protein complexes and used to make haemoglobin and myoglobin [135]. In this case the biotransformation also leads to bioclearance of the NPs. Other biomodification routes include lysosomal degradation of NPs as a result of the low pH in the lysosomes coupled with their high enzymatic composition and indeed their role as the “dustbin” of the cell.

The composition of NPs can be conceptually divided into the often inorganic core; the engineered surface coating comprising of the ligand shell and optionally also bio-conjugates; and the corona of adsorbed biological molecules [136]. Empirical evidence shows that all three components of NPs (core, shell and corona) may degrade individually in vivo and can drastically modify the life cycle and biodistribution of the whole heterostructure, such that the biodistribution and fate of each sub-component would need to be analyzed individually for regulatory and nanomedical approval purposes [136]. Approaches to do that, based on stable isotope and radiolabelling of core and shell separately are emerging, with differential biomodification process demsonstrated for polymer-coated FeO_x NPs [136] versus Au NPs [137]. Thus, approaches to assess biomodification, bioprocessing, and bioclearance are emerging, and these are intrinsically linked. The co-evaluation of core and shell need to be determined on a case-by-case basis until predictive models can be developed.

A range of studies have looked at the correlation between NP properties such as size and surface charge with uptake, biodistribution and bioclearance (defined here as removal from specific organs (e.g. the lung or gut) or from the organism overall. Blanco, Shen and Ferrari looked at effect of size (<5nm, 20-150nm and > 150nm), shape (20-150nm spheres, rods and discs) and surface charge (20-150nm spheres with negative, neutral or positive) on where the NPs accumulated [138]. The findings indicated that the kidney was the main target organ for <5nm NPs, whereas for the 20-150nm spheres, positive charge correlated with enhanced accumulation in liver and spleen. Discs appeared to show most accumulation, collecting in liver, spleen and lungs, with the order of accumulation for shapes being discs > rods > spheres [138]. There is clear evidence that each of these parameters also influences the nature of the biomolecules bound [139–141], and that different coronas lead to different biodistributions of NPs in vivo, as indicated above and demonstrated by Wang et al. [142]

It becomes clear from the snapshot of studies presented above that it is very difficult to untangle the 5Bs – they are interrelated and inter-dependent, but a clear understanding of each, and their combined impact, is vital for regulatory certainty. Fig. 6 provides an overview of our current conceptual understanding. There is a clear need for parallel in vitro and in vivo studies in the short term in order to untangle the pathways and mechanisms involved, with the in vivo studies in particular providing important insights into the final biomolecule coronas following in vivo biodistribution and/or during bioclearance [143]. While there is a strong drive to reduce reliance on animal testing, this can only be achieved once in vitro and in silico methods have proven to be predictive of in vivo health outcomes. Finally, it is clear that the adsorbed biomolecules play a central role in each of the processes underpinning NP bioavailability, biopersistence, bioprocessing including biodistribution, biomodification and bioclearance. Taken together, these studies suggest that there is still a major gap from fundamental science to regulatory relevance, but that progress is being made, and that the biomolecule corona may provide important insights and lead to the potential for predictive toxicology based on the bioprocessing signalling predicted from maps or fingerprints of NP-associated biomolecules. The formation of the biomolecule corona around NPs also raises the question of what is the relevant “form” to assess in regulation (e.g. the evolving NP-corona complex versus the pristine NP), especially as numerous studies show that bare particles are both more toxic, and rapidly acquire a biomolecule corona from their surroundings.

Future directions for research into the 5Bs include development of predictive models for NP corona formation and composition, and for corona influence on cellular attachment and uptake, biodistribution, bioprocessing and bioclearance, and as well as elucidation of key initiating events and the subsequent adverse outcome pathways related to biomodification resulting from exposure to NPs. From a biomedical perspective, significant research directions include understanding and controlling targeting of NPs to the desired location and the role of the biomolecule corona in driving this, as well as understanding and predicting the impact of the biomolecule corona on drug release rates and reducing off-target effects arising from sub-populations of the NPs being bioprocessed or biodistributed differently to the ideal. Significant effort is needed to elucidate the in vitro – in vivo correlations, especially given the very different biomolecule concentrations typical of each (10% versus >80% biomolecules, respectively), and the different biomolecule compositions in discrete tissue compartments in vivo, and the consequences of this for NP bioavailability, bioprocessing and bioclearance.

Immunity and systemic responses

Nanosafety assessment needs simple and affordable, but at the same time robust and meaningful, assays. Readouts reflecting immune activation are in this context pursued by many groups [144]. This is reasonable, since the immune system has evolved to recognize non-self and to decide whether a defensive action is appropriate. In addition, immune cells are concentrated at the potential routes of entry for pathogens, which are the same routes by which NPs enter the body. This perspective will discuss what information can be gained when focusing on systemic rather than local immune responses.

Entry of a potentially dangerous non-self entity stimulates local reactions. If danger signals are sensed, a local inflammation ensues. The symptoms are familiar: Reddening and swelling (edema formation due to immigration of immune cells, stimulated by locally produced chemotactic factors), heat (from increased blood flow, to limit bacteria and viruses which often have a narrow temperature optimum), as well as pain and impaired function (limiting movement to avoid further injury and support wound healing). However, most contacts with non-self do not result in defence, due to the lack of danger signals. Still, the immune system will respond: Either with homeostatic fluctuations of no further consequence, or with the development of tolerance via several active mechanisms. The majority of immune responses indeed result in tolerance, which can be a source of misinterpretation. Nanosafety assessment requires distinguishing whether a response is merely adaptive or truly defensive, which can be addressed by the choice of endpoints. Most studies on immune effects of NPs have addressed the inflammatory (innate) immune response, with respect to local reactions [145–147].

Parameters that are characteristic for local responses are well detectable in many professional immune cells, but also in tissue cells which in vivo produce alarm signals to attract immune cells into a potentially threatened site. A good marker for local response is activation of the transcription factor NF-kB, which indicates cell stress that occurs during immune stimulation [45, 148]. The chemokine IL-8 is another example: It is produced very early and can be defined as a relatively unspecific alert signal, which indicates local inflammation that can be extreme enough to induce cytotoxicity [149–152]. These markers are readily measured and convenient for simple and affordable tests, but their use with not well characterized materials and cells requires care to confirm that an observed response is indeed defensive. In this respect, NPs present the challenges of heterogeneity and of batch-to-batch variation. Readouts like NF-kB and IL-8 are popular since they can be readily measured in cell lines and primary cells on the transcriptional and the protein secretion level, using methods like qRT-PCT, reporter genes and ELISA. However, , there is a risk of false positives when a normal homeostatic fluctuation in response to a stimulus is mistaken for an indication of danger to the body. Even worse, inflammation may be due to contamination, most commonly with the ubiquitous bacterial compound LPS [153] .

One way to deal with this problem is to look for parameters indicating systemic responses. For example, IL-1 is a major pro-inflammatory, fever inducing cytokine produced by several immune cell types, so detecting substantial IL-1 secretion would suggest a systemic inflammation which is certainly not indicating tolerance [154]. The related cytokine IL-18 shares many functions of IL-1 (but not fever induction) and offers the advantage that it is produced by numerous cell types [155]. Such readouts indicate a systemic inflammation that evolves from a local one: Think about flu that after a while affects the whole body, despite a very local occurrence of virus. The advantage of systemic inflammation markers is that they more clearly indicate a defensive reaction, thus immune mechanisms have assessed a stimulus as being dangerous.

Inflammation is often considered to be synonymous with innate immunity. This is an evolutionary ancient package of defensive mechanisms, which has the attractive feature that similar readouts can be made in invertebrates, offering opportunities to directly link human toxicology and ecotoxicology. In contrast, adaptive immunity has fully developed first in teleost fish and is thus limited to vertebrates. Adaptive reactions are always systemic, since they require the interaction of several cell types (especially antigen presenting cells, T-cells and B-cells) and involve secondary immune tissues, mainly lymph nodes and spleen. As in the case of innate immunity, a response is not the same as a defensive action. T-cells may be activated to proliferate and differentiate, but if they should be of the T_reg type (regulatory T-cells), they will produce immunosuppressive cytokines and thus promote tolerance.

Adaptive immune responses can develop in three directions, of which the “default” is tolerance. Type 1 and type 2 responses are defensive, the first being directed mainly against bacteria and viruses, the second mainly against macroparasites. Both are distinguished by their cytokine pattern, especially for the cytokines produced by T-cells. IFN-α and IL-4 are among the tell-tale markers for type 1 and type 2 responses, respectively. A functional definition can be made via the isotypes of antibodies that are produced by B cells. IgG₁ is the most prominent antibody type in blood and it increases substantially during type 1 responses. Since adaptive reactions develop more slowly than innate ones (days vs. hours), it makes sense that type 1 effectors like IgG₁ interact productively with innate immune mechanisms. Assessing type 1 responses is more challenging than testing for inflammation and cell stress, since a full response can be mimicked only in co-culture systems rather than in single cells. Some NPs have been shown to influence the development of antibodies, even though they are themselves only rarely recognized by antibodies [156].

Type 2 responses are associated with parasites, but are for many people more familiar as a pathophysiological response in allergic diseases. IgE antibodies mediate these responses. They have very low levels in serum, but are bound to high-affinity receptors on the surface of effector cells (basophils, mast cells, eosinophils). So far, no case has been described where NPs act as allergens [157], but it has been shown that binding of allergens to NPs can enhance their allergenicity [158].

So far, most studies have investigated immune effects of NPs as single agents. In the future, we can expect an increasing number of studies that treat NPs as one component in a complex exposure situation, which corresponds more closely to real life. In addition, NPs can have effects on the development of a systemic antibody-mediated response, but it is not well known under which circumstances and to which extent this takes place in the body. Nanomedicine can address these issues more easily, since it deals with high concentrations and controlled exposure. Developing methods that allow considering adaptive immune reactions for risk assessment is at present a challenge. The effects of co-exposure to NPs and to stimuli that induce either systemic type 1 or type 2 responses will increasingly become a subject for nanoimmunotoxicology research aiming at risk assessment.

Non-Specific Effects of Particles

As non-specific effects of particles, mixed dust pneumoconiosis and response of autonomic nervous system to exposure to particles are discussed.

Particles and the Developing Body

Understanding the developmental toxicity of NPs is vital because exposure to fine particulate matter during gestation increases the risk of low birth weight in the child [164], which is associated with health and disease later in life [165, 166]. Recent clinical cohort studies suggest that prenatal and neonatal exposure to fine and ultrafine particulate air pollution is associated with an increased risk of developmental brain disorders, such as autism spectrum disorder and schizophrenia, in offspring [167–169]. It is important to understand the mechanism of action underlying the effect of NPs on the developing body in order to reduce the toxic risk of these atmospheric ultrafine particles and to promote the safer use of engineered NPs for future generations.

We start to have some understanding of the mechanisms underlying the potential hazards of fine and ultrafine particles to brain and behaviour. The effects of particle exposure can be studied using brain perivascular histopathology [170]. In particular, the expression levels of proteins associated with astrogliosis, e.g. glial fibrillary acidic protein, may be useful as a sensitive and quantitative marker of maternal exposure to low doses of NP for prediction of their developmental toxicity [171]. There seems to be potential protective effects of antioxidants on the brain perivascular abnormality (astrogliosis) of offspring mice whose mothers were exposed to carbon black NPs. One of the anti-oxidants, N-acetyl cysteine, partially suppressed the astrogliosis in the brain of offspring induced by maternal exposure to carbon black [172]. There are also developmental effects of perinatal exposure to experimental secondary organic particles as prepared by oxidation of diesel exhaust particles with ozone. Neonatal mice exposed to secondary organic particles demonstrated a decrease in social behaviour with down-regulation of estrogen receptor-β and oxytocin receptor in the hypothalamus [173]. Ming-Wei Chao et al. reported an increase in reactive oxygen species and several cytokines in the amniotic fluid, and changes in microRNA expression profile in the cerebral cortex and hippocampus of foetal brains in rats following exposure to PM_2.5 [174].

Recently, the potential use of zebrafish (Danio rerio), Drosophila, and chicken embryos has been investigated as alternative methods for high-throughput screening of the developmental toxicity of NPs [175]. The zebrafish model, in which hatching rate, developmental malformation of organs, genotoxicity, immunotoxicity, abnormal behaviour, and neurotoxicity can be measured, is useful in the field of toxicology and biomedical research to evaluate the reproductive and developmental toxicity of NPs [176]. The advantages of the zebrafish model are their small size, high reproducibility, quick development, and transparency of the embryo [177]. The transparency of the embryo enables observation of all the cells from early larval stages and facilitates real-time imaging of NP distribution in vivo during the embryonic development [178]. In addition, their genetic information is accumulating rapidly by genome sequencing [179]. This model has also allowed comparative analysis of the developmental toxicity of NPs with different properties, such as size [180, 181], chemical composition [182, 183], and surface modification [184]. On the other hand, the chicken embryo model has an advantage that the direct effect of NPs on the embryonic development can be evaluated by removing the indirect effect mediated by maternal factors easily, as reported for the results of developmental toxicity of TiO₂-NP [185] and carbon NPs [186–188]. In addition, the use of Drosophila, which has a short life span and 77% of the human disease genes [189], has just started in terms of developing cost-effective high-throughput screening methods for assessment of the developmental toxicity of NPs [190]. These animal models may provide rapid hazard assessment techniques to facilitate regulation and ensure safer NPs reach the market thereby protection future generations.

Previously, the translocation of NPs from maternal to foetal circulation, with ultimate deposition in the offspring’s body, was first reported in an experimental model where TiO₂ NPs were subcutaneously injected into pregnant mice [191]. Experiments with an ex vivo placental perfusion model showed an inverse correlation between the translocation of nano- and submicro-particles (in this case, polystyrene beads) and particle size [192]. Recently, the dose- [193] and size-dependency [194] of NP translocation from the mother to foetus was also reported in in vivo studies. Overall, NP translocation through the placenta is likely dependent on particle surface modification [195, 196], chemical composition [197], and the timing of the exposure during pregnancy [196]. Because the mass-based translocation ratio is not high, the detection of these particles is dependent on the analytical techniques used. It is important to understand not only the direct effect of NPs on the foetus, but also the indirect effects mediated by circulating cytokines or other secondary messengers generated from oxidative stress and pulmonary inflammation in the maternal lungs [198–202] [174], placental dysfunction [199], or genotoxicity [203, 204].

The reproductive [191, 205–210] [211], immune [200, 212, 213], and central nervous systems have been investigated as potential targets of maternal NP exposure. In particular, the effect of this exposure on the offspring brain is getting better understood. While evaluation of developmental neurotoxicity via histopathology [170, 214, 215], and determination of DNA methylation level [216], monoamine level [217–220], and expression profile of mRNA [221, 222], X-chromosome inactivation factor [223] and microRNA [174] is valuable, evaluating the neurobehavioral changes in children after prenatal NP exposure is also important in understanding the developmental neurotoxicity of NPs. Previous studies have shown that exposure to NP-rich diesel exhaust or concentrated ambient particles (CAPs) during prenatal and/or neonatal period altered spontaneous locomotor activity level [217, 224, 225], decreased novel object recognition [226] and motor coordination [218], and increased autism-like repetitive and impulsive behaviours [167, 224]. Moreover, maternal inhalation studies showed that exposure to nanoparticulate TiO₂ during the foetal period decreased visits to, and duration of, stay in the central zone during evaluation with the open field test in mice [198] and impairments of working or short-term memory and initial motivation in rats [227]. A decrease in spontaneous locomotor activity in a novel environment was also reported in mice whose mothers were exposed to carbon black NPs [201]. The sex difference in response to nanoparticle exposure during the fetal/perinatal period is likely important [167, 198, 201, 225, 228]; however, further investigation is needed to clarify the mechanism of the difference.

Histopathology suggests that brain perivascular cells, including perivascular macrophages and their surrounding astrocytes, have an important role in clearing waste from the brain parenchyma [229, 230]. They can serve as an extremely sensitive marker of maternal exposure to low doses of NPs for prediction of their developmental toxicity [170]. The histopathological changes were confirmed with decreased number of parvalbumin-positive interneurons following a maternal inhalation study [231]. Sensitive and quantitative endpoints evaluating the developmental impacts of maternal NP exposure are necessary to facilitate the risk assessment and hazard identification of NPs [232]. Future work will hopefully clarify the mechanism of developmental toxicity of NPs, and aid in the development of preventive strategies against intended and unintended NP exposure.

Particles in the natural environment and links to human health

Lessons from the past for future toxicological studies on big, small and variously shaped particles

Risk management and governance for particles and fibres

Risk governance serves as an organizational framework through which the critical elements of risk assessment, management, and communication are applied to the scientific and technological solutions for local, regional, and global issues. Through risk governance, the beneficial and adverse impacts of such solutions are understood, and using principles of cooperation and participation, effective policy decisions are made. Over the last 20 years, as the concept of risk governance has evolved, its scope has become more comprehensive, the number of stakeholder communities larger, and its practical application daunting Fig. 7 [416]. Also for nanomaterials (including NPs) risk governance, a series of risk governance strategies have been developed [417, 418]. In assessing scientific and technical products, a balance between innovation – the process that provides a new product or solution that is technologically possible and viable in the market place – and sustainability – the process that considers what is bearable and equitable for society, the environment, and the economy across the life cycle of the solution – must be struck. The challenges of stakeholder involvement in risk governance decision making are complicated by difficulty identifying and engaging the appropriate stakeholder communities, the limited coordination between government and non-government communities, and the inability of governments to agree on the principles and elements of risk governance. Perhaps the biggest challenge is the identification of a convening authority (or authorities) with the multidisciplinary expertise, neutrality, and objectivity needed to convene communities and governments within, and across, nations for consensus decisions. As one analyzes the complexity and challenges represented by this approach, one might indeed ask if the scope and organization of risk governance has made it too big to succeed and if a need has arisen to create a more focused, adaptive, and practical risk governance approach.

At the core of risk governance lays the continuous assessment, appraisal, evaluation and management of the risks. This process translates governance goals into concrete strategies for dealing with the risks posed by particles and fibres. Approaches to deal with hazardous dusts focused in the past on hazards and appropriate controls measures [419]. This has evolved to approaches that appraise the risk, which is the product of hazard and exposure, before defining strategies to manage this risk (e.g. [420]). To understand the risk of particles and fibres, it is necessary to collect basic information on the physico-chemical properties such as chemical composition, size, and shape, biopersistence, and reactivities; testing results such as acute and chronic toxicity; and dosing information such as bio-kinetics, exposure levels, frequencies and duration of exposure to consumers, workers and the general population. In a second step, these data are combined to appraise the risk in terms of people harmed, disabled or killed, and related metrics such as disability-adjusted life years or cost to victims, companies, insurance and society. Next, risk management measures are defined that aim to reduce the risk to acceptable levels. It is good occupational hygiene practice to further reduce risks that are easy to reduce. Considerable attention was given recently to nanomaterial and in particular NP risks and the knowledge gaps that science has to fill [5]. In addition to the challenges faced with controlling the risk from larger particles and fibres, NPs bring along two special challenges: first, materials often behaves differently at the nanoscale lending them different physico-chemical properties. Second, NPs were found to translocate (as discussed above) to different target organs compared to micron sized particles. Both of these challenges imply that data obtained in the past for larger particles of these materials may no longer be valid for the nanoform and a series of novel strategies has been advanced to identify properties that best predict NPs’ risk potential [421, 422]. Taken together, modern risk governance and risk management theories propose assessing risks in a comprehensive framework that aims at reducing the risk and the uncertainty related to it. In such a framework, one needs to understand the likely exposure levels and how they relate to release from products and production processes, how these concentration ranges translate into dose-response distribution functions and how they can be combined with dose-response functions and uncertainties of the NPs’ hazards. By focusing on optimizing the risk rather than addressing exposures and hazards individually, considerable gains are expected both in risk reduction as well as in terms of the time and cost required to reach acceptable risk levels.