Lessons learned from research on air pollution and other particles in the toxicology of nanomaterials and vice versa.

Abstract

The increasing number of new nanotechnological applications calls for the development of a test paradigm for nanomaterials (NMs) that is relevant to both pristine particles, in their intended use, and as waste products. NMs have an enormous potential for variation in physicochemical properties, including differences in size, shape, aggregation, chemical composition, surface modifications, and crystal structure. Thus, knowledge on how to evaluate the potential hazards associated with exposure to NMs, general mechanisms of toxicity and how to group and rank NMs according to the toxicity is urgently needed. The International Agency for Research on Cancer (IARC) monographs are widely acknowledged as an authoritative source of evaluation of carcinogens. IARC has repeatedly issued monographs on various types of water-insoluble materials, including asbestos fibers and crystalline silica. The mechanisms of action of carcinogenicity by exposure to particle and fiber have been elucidated over a relatively long period of time. The IARC evaluation of man-made vitreous fibers in 2001 highlighted the biopersistence, incomplete phagocytosis, persistent inflammation and production of reactive oxygen and nitrogen species as important mechanisms of action [IARC Working Group on the Evaluation of Carcinogenic Risks to Humans, 2002]. Two years later, IARC evaluated some “particulate toxicants,” including cobalt with or without tungsten carbide in hard metals. The review on mechanistic aspects considered genoxicity to arise from Fenton-type reactions (i.e., production of reactive oxygen species by transition metals) and inhibition of DNA repair by cobalt (II) ions. Based on limited evidence for carcinogenicity in humans and sufficient evidence in animals for some types of cobalt materials, the overall evaluation settled on cobalt with tungsten carbide to be probably carcinogenic to humans (Group 2A). A majority decision of the working group members advocated for the need for either sufficient evidence in humans or strong mechanistic evidence to upgrade to a Group 1 carcinogen [IARC Working Group on the Evaluation of Carcinogenic Risks to Humans, 2006]. Although the production of reactive oxygen species and inflammation were described as mechanisms in earlier monographs, a clearly conceptual framework of lung carcinogenesis in rats was only later depicted in the monograph on carbon black, titanium dioxide and talc, which were evaluated in 2006 as examples of chemically inert and poorly soluble particles [IARC Working Group on the Evaluation of Carcinogenic Risks to Humans, 2010aa]. As outlined in the general remarks of the monograph, the working group did not evaluate the ultrafine or nano-form of these particles, although it was noted that the mechanistic studies suggested that smaller particles may be more effective than larger particles in inducing toxic effects. The working group concluded that there was inadequate and sufficient evidence for carcinogenicity in humans and animals for carbon black and titanium dioxide, respectively. However, there was not sufficient mechanistic evidence to warrant a classification higher than Group 2B, which was primarily based on uncertainty pertaining to whether the mechanic information is likely to operate in humans. Recently, the volume 100 series of IARC evaluations reviewed a number of previously classified Group 1 carcinogens; volume 100C is relevant in nanotoxicology because it reviewed a number of metals, fibers and dusts. The outline of the mechanism of carcinogenicity of asbestos fibers included frustrated phagocytosis, leading to inflammatory cell recruitment and activation, production of reactive oxygen species, genotoxicity (DNA damage, mutations, and chromosomal alterations), and activation of oncogenes or inactivation of tumor suppressor genes. In a similar manner, the proposed mechanism of carcinogenicity of crystalline silica in rats included inflammation and oxidative stress central intermediate steps. Figure 1 shows this current mechanistic concept for poorly soluble particles and fibers. Mechanistic concept of how poorly soluble particles and fiber cause cancer by genotoxicity in a pro-inflammatory and pro-oxidant milieu. The above concept has also been used in monographs from 2006 on complex mixtures such as indoor emissions from household combustion of coal (Group 1) and biomass (Group 2A) [IARC Working Group on the Evaluation of Carcinogenic Risks to Humans, 2010bb], and in the 2012 monographs on engine exhaust from combustion of diesel (Group 1) and gasoline (Group 2B) [IARC Working Group on the Evaluation of Carcinogenic Risks to Humans, 2013]. Air pollution – a mixture of mixtures – was also classified as carcinogenic to humans (especially the particulate fraction) (Group 1) in 2013 [Benbrahim-Tallaa et al., 2012]. Complex mixtures also pertains to NM exposure, although it has only been sparsely investigated. For instance, one study investigated the genotoxic potential of pristine pigments and sanded dusts of conventional and NM-containing paints [Saber et al., 2012a, 2012b]. In 2014, IARC evaluated the carcinogenic risk to humans of exposure to carbon nanotubes (CNT), which can be seen as the first real attempt to assess the impact of exposure to engineered NMs on cancer risk [Grosse et al., 2014]. Interestingly, there was sufficient evidence in animals for a specific type of multi-walled carbon nanotubes (MWNT-7), whereas there was insufficient information about mechanistic aspects. Therefore, there was not sufficient mechanistic evidence to warrant a classification higher than Group 2B. There was limited evidence for the two other types of MWCNTs with dimensions similar to MWNT-7 (Group 3), and inadequate evidence for SWCNTs (Group 3). Ironically, there was a relatively higher abundance of data on genotoxic mechanisms of action for these types of CNTs. Figure 2 depicts the distribution of IARC classifications, including certain types of materials for which particle and fiber toxicity has been used in the evaluations. Interestingly, there are a number of agents that have been classified mainly by sufficient evidence in epidemiological studies (Group 1 agents), whereas CNTs and poorly soluble particles have only merited Group 2B or 3 classification. This suggests that the mechanistic data is the Achilles heel in the classification and there is paucity for mechanistic studies on particles and fibers. Thus, mechanistic studies will be of paramount importance to the classification of NMs because high-quality epidemiologic studies are probably not going to be available in the immediate future. Distribution of agents that have been evaluated by The International Agency for Research on Cancer (IARC) and specific classifications of certain types of poorly soluble particles and fibers. Genotoxicity is a crucial endpoint in safety testing as it assesses potential mutagenicity and clastogenicity, which has implications for risks of both genetic disease and carcinogenesis. Recently, a research strategy was proposed for the development of an intelligent testing strategy for NMs [Stone et al., 2014]. The research strategy is based on several steps. The first step is identification of the biological mechanism underlying the toxic effect and the physicochemical properties of the NMs that drive the toxic effects. The next step is to design in vitro and high-throughput screening tools that target these key biological processes. These two steps will allow grouping and ranking of NMs and NM-containing products according to their inherent toxicity. The third step is to develop robust in silico approaches for risk assessment. Another approach towards the development of alternative testing strategies and high throughput methods for hazard assessment of NMs has been pursued in the NanoTEST project (www.nanotest-fp7.eu;) [Dusinska et al., 2009; Dusinska and Tran, 2014]. This research focused on NMs potentially used in nanomedicine, taking advantage of the fact that for biomedical NMs the exposure is known, NMs enter the body and the interaction of NMs with cells and tissues is inevitable. Cytotoxicity, oxidative stress, immunotoxicity, and genotoxicity were investigated in various cell culture models, representing eight different organs (blood, vascular system, lung, brain, liver, kidney, gastrointestinal system, and placenta). All in vitro studies were harmonized, using NMs from the same batch, and identical dispersion protocols, exposure time, concentration range, culture conditions, and time-courses. The in vitro methods were critically evaluated, and where appropriate, standard methods were adapted [Dusinska et al., 2009]. The suitability of human and mammalian primary cells and cell lines derived from blood, vascular/central nervous system, liver, kidney, lung, and placenta for the assessment of NM genotoxicity (DNA strand breaks and oxidized DNA lesions) was investigated by Cowie et al. [2014]. The results from the statistical evaluation showed that all of the cell types can be used to assess the genotoxic potential of NMs, but with different sensitivities. This work also assessed the effect of changes in experimental conditions on the toxic impact of NMs in different cell culture models. The results of these studies have been used to generate recommendations for a suitable and robust testing strategy that can be applied to new medical NMs as they are developed. The recommendations include needs for standardized characterisation of NMs, measurement of NM-uptake and for each type of toxicity endpoint at least two different methods. The panel of NMs dispersed in medium showed that many characteristics (composition, size, coatings, and surface properties) interfere with a range of in vitro cytotoxicity assays [Guadagnini et al., 2014]. In a recent editorial, Poland et al. [2014] advocated for toxicological testing with a focus on reproducibility. Here, the authors emphasized the need for detailed and careful characterization of the tested NMs, and rigorous statistical testing of the results. Furthermore, the authors emphasized the need for the researchers to be critical of their own hypotheses and results, and to be the first to try to disprove the proposed hypothesis. Safety assessment of NMs depends on knowledge of their effects at different levels – cells, organs, animals, and humans. Such knowledge will help with the introduction of guidelines for the safe production, use, and disposal of NMs. We, the guest editors of this Special Issue on nanotoxicology for Environmental and Molecular Mutagenesis, agree to the need for robust and standardized methods in nanotoxicology and the need for comprehensive material characterization as an integral part of nanotoxicology. It is also important to critically evaluate the quality of the utilized assays and their predictive value for the hard endpoints of interest such as cardiovascular diseases, cancer or neurodegenerative diseases [Møller et al., 2011]. These diseases have a protracted sub-clinical course before diagnosis in humans. Insight into mechanisms of NM-generated health effects is therefore a pivotal step in the development of testing strategies for hazard assessment of NMs. The articles in this special issue of Environmental and Molecular Mutagenesis are contributions that will feed into future evaluations of engineered NMs by agencies like IARC, but they also contribute to the consorted research aim to develop a testing strategy for NMs. Although the goal is the same – to protect workers and consumers – the approach used to evaluate the risk of carcinogenicity for a specific engineered NM differs substantially from those used to develop testing strategies. The approach to evaluate a specific type of NM will put substantial emphasis on mechanistic steps that are close to the hard endpoint (e.g., mutations in oncogenes or tumor suppressor genes provide stronger mechanistic evidence than inflammation). However, the assessment of mutations in relevant oncogenes or tumor suppressor genes is costly and it is not suitable as a routinely applied high throughput screening test for large numbers of samples. The latter would typically strive to develop assays that assess a specific endpoint in the presumed causal pathway from exposure to disease endpoint (e.g., measurement of DNA damage). The strive toward pathway-based assays is clearly represented through the assessment of redox status in NM-exposed cells and animals, which is an endpoint that is suitable as a high throughput assay in target tissues [Kermanizadeh et al., 2015]. In this area, [Karlsson et al., 2015] in this issue provide a convincing analysis that the comet and micronucleus assays display high concordance, despite the fact that they detect different mechanisms of genotoxicity. Nevertheless, there are also dark clouds on the horizon as shown by Møller et al. [2015], who conclude that using “poor” assays is associated with a higher likelihood of finding (or reporting) genotoxic effects of NMs. This is a counter-intuitive conclusion, based on published literature, but it is also a reminder that we should strive to use the best methods available. Moreover, it is broadly indicative of one of the pitfalls of publishing, where positive results are sometimes viewed as more exciting (and publishable) than null effect findings. Although focus has been drawn to the effects of engineered NMs, concerns over the health effects related to “conventional” exposure of particles still linger. For instance, in the wake of the IARC evaluation of diesel engine exhaust, concern was expressed as to the classification being based on old technology and fuels [McClellan et al., 2012]. Biodiesels have received increasing focus as alternatives to fossil fuels, although biodiesel combustion may generate particles that are smaller than conventional diesel fuels [Hemmingsen et al., 2011]. However, biodiesel may also pose increased risk of reproductive effects as reported by Kisin et al. [2015] who observed that exposure to biodiesel was associated with testes interstitial edema, degenerating spermatocytes, dystrophic seminiferous tubules with arrested spermatogenesis, and increased sperm DNA fragmentation, possibly mediated by oxidative stress (increased lipid peroxidation products and depletion of glutathione) and inflammation mechanisms. Careful hazard assessment of different nanoparticles provides an important first step in the attempt to group and rank nanoparticles. Valdiglesias et al. [2015] have reviewed the current knowledge on iron oxide nanoparticle associated toxicity showing little consensus among the different studies that could be due to the different surface coatings and particle size of the investigated nanomaterials. Migliore et al. [2015] review the evidence of neurodegenerative effects of NM exposure. In their review, they conclude that NMs may contribute to the onset and progression of several human neurodegenerative diseases due to the toxic properties of many nanosized particles, but nanotechnology can take advantage of the physicochemical properties of NMs for delivery of either diagnostic or therapeutic compounds to the site of the affected regions in the brain. The papers by Guichard et al. [2015] and Tarantini et al. [2015] describe the careful and detailed toxicological testing of synthetic amorphous silica nanoparticles in rats by all relevant routes: oral, pulmonary, and intravenous injection using Organisation for Economic Co-operation and Development (OECD) standard materials. These authors report no genotoxicity following oral and pulmonary exposure to synthetic amorphous silica nanoparticles. Thus, these articles contributes high quality in vivo data to be used not only in the risk assessment of amorphous silica, but also in the overall understanding of particle-induced genotoxicity. Catalán et al. [2015], Jackson et al. [2015], and Kansara et al. [2015] explore the mechanism of particle and fibre-induced genotoxicity in vitro. Jackson et al. [2015] screened 15 different carefully characterized carbon nanotubes for genotoxity reporting little cytotoxicity and genotoxicity, whereas Kansara et al. [2015] explored the potential molecular mechanisms of titanium dioxide-induced double stand breaks in vitro. Catalán et al. [2015] assessed the genotoxic and immunotoxic response to nanocellulose crystals, reporting that the studied nanocellulose crystals did not induce genotoxicity or inflammation. The question of grouping and ranking is also addressed by Halappanavar et al. [2015], who compare global transcriptional profiles in mouse lung tissues following pulmonary exposure to different titanium dioxides and sanding dust from titanium dioxide-containing paints. The information is used to investigate which physicochemical properties of TiO2 nanoparticles determine transcriptional response. The study indicates that although the underlying mechanisms leading to inflammation and acute phase response suggest a generalized mode-of-action for all TiO2 nanoparticles, a combination of smaller size, large deposited surface area and surface amidation are main contributors to toxicity. In conclusion, studies on asbestos fibers, silica, air pollution and diesel exhaust particles have provided strong evidence of an association between exposure and cancer endpoints, with biopersistence, persistent inflammation and genotoxic events as important intermediate steps. Progress toward assessment of cancer risk of particle exposures has entailed studies on carbon black, TiO2 and CNTs. There is much still to be done regarding hazard identification and risk assessment of NMs, but this issue presents new knowledge on toxicity of NMs and highlights problems that need to be solved to develop a comprehensive strategy for the risk assessment of NMs. Moreover, knowledge gained here on the mechanism of action of NMs may prove relevant for an improved understanding of the toxicology of particles emitted from combustion of modern fuels in new technology engines, which will be relevant in future studies on air pollution particles. Peter Møller Institute of Public Health Section of Environmental Health University of Copenhagen Copenhagen K, Denmark Maria Dusinska Norwegian Institute for Air Research Health Effects Laboratory Department of Environmental Chemistry Kjeller, Norway Ulla Vogel* National Research Centre for the Working Environment Copenhagen, Denmark; Department of Micro- and Nanotechnology Technical University of Denmark Lyngby, Denmark