The New Toxicology of Sophisticated Materials: Nanotoxicology and Beyond

Abstract

It has long been recognized that the physical form of materials can mediate their toxicity—the health impacts of asbestiform materials, industrial aerosols, and ambient particulate matter are prime examples. Yet over the past 20 years, toxicology research has suggested complex and previously unrecognized associations between material physicochemistry at the nanoscale and biological interactions. With the rapid rise of the field of nanotechnology and the design and production of increasingly complex nanoscale materials, it has become ever more important to understand how the physical form and chemical composition of these materials interact synergistically to determine toxicity. As a result, a new field of research has emerged—nanotoxicology. Research within this field is highlighting the importance of material physicochemical properties in how dose is understood, how materials are characterized in a manner that enables quantitative data interpretation and comparison, and how materials move within, interact with, and are transformed by biological systems. Yet many of the substances that are the focus of current nanotoxicology studies are relatively simple materials that are at the vanguard of a new era of complex materials. Over the next 50 years, there will be a need to understand the toxicology of increasingly sophisticated materials that exhibit novel, dynamic and multifaceted functionality. If the toxicology community is to meet the challenge of ensuring the safe use of this new generation of substances, it will need to move beyond “nano” toxicology and toward a new toxicology of sophisticated materials. Here, we present a brief overview of the current state of the science on the toxicology of nanoscale materials and focus on three emerging toxicology-based challenges presented by sophisticated materials that will become increasingly important over the next 50 years: identifying relevant materials for study, physicochemical characterization, and biointeractions.

In 1990, two consecutive articles appeared in the Journal of Aerosol Science asking whether inhaled particles smaller than 100 nm in diameter elicit a greater than expected pulmonary response (Ferin et al., 1990; Oberdörster et al., 1990). On a mass for mass basis, nanometer-scale particles of TiO₂ and Al₂O₃ elicited a significantly greater inflammatory response in the lungs of rats compared with larger particles with the same chemical composition. The two studies were at the vanguard of research challenging long-held assumptions that response to particulate exposure can be understood in terms of chemical composition and suggested unusual biological activity associated with nanometer-scale materials. Fourteen years later, this growing field of research would be formalized as the field of “nanotoxicology” (Donaldson et al., 2004).

The size-specific effects observed by Oberdörster, Ferin and colleagues were attributed to an increased rate of interstitialization of nanometer-scale particles in the lungs. Oberdörster et al. concluded, “Phagocytosis of particles in the alveoli counteracts the translocation of particles into the interstitial space. Alveolar macrophage death or dysfunction promotes translocation from alveoli inter interstitium. Particles of about 0.02–0.03 μm in diameter penetrate more easily than particles of ∼0.2–0.5 μm. Small particles usually form aggregates. Their aerodynamic size determines the deposition in the airways. After deposition, they may deagglomerate. If the primary particle size is ∼0.02–0.03 μm, deagglomeration may affect the translocation of the particles more than for aggregates consisting of larger particles” (Oberdörster et al., 1990).

This simple statement outlined two emerging aspects of materials that potentially mediated their toxicology: particle size and dynamic behavior. In follow-up studies, further associations between material physicochemistry and effects were uncovered—most notably the role of particle surface area in mediating pulmonary toxicity. Using TiO₂ samples comprising of two distinct sizes of primary particles, Oberdörster et al. showed that, while inflammatory responses following inhalation in rats depended on particle size, normalizing by surface area led to a size-invariant dose-response function (Oberdörster, 2000). With surface area as the dosemetric instead of the more conventional mass concentration, Maynard and Kuempel (2005) and others showed that a range of insoluble materials typically classified as “nuisance dusts” followed a similar dose-response curve for pulmonary inflammation in rats. However, more chemically active materials such as crystalline quartz demonstrated a markedly different dose-response (Maynard and Kuempel, 2005).

This early research was driven by occupational aerosol exposures and concerns that the hazards associated with fine dusts ranging from welding fume to metal and metal aerosol powders were not predictable from the chemical composition of these materials alone. What began to emerge was an understanding that the physicochemical nature of inhaled particles was more relevant than previously thought in eliciting a response and that materials with a nanometer-scale biologically accessible structure (whether they were discrete nanometer-scale particles or had a nanometer-scale surface structure, as in the case of aggregates of nanoparticles) had the potential to show previously unrecognized biological behavior. That this new research on what were termed “ultrafine aerosols” came out of occupational toxicology is perhaps not surprising, given the field's long history of addressing hazards associated with exposure to aerosol particles with varying sizes, shapes, and compositions (Maynard, 2007a).

Although research into occupational exposure to ultrafine aerosols was developing in the 1990’s, environmental epidemiology studies were beginning to uncover associations between ambient aerosol particle size and morbidity and mortality. Starting with the six-cities study (Dockery et al., 1993), evidence emerged for ambient particles approximately smaller than 2.5 μm (PM 2.5) having an elevated impact on human health (Pope, 1996; Schwartz and Morris, 1995; Schwartz et al., 1996). As small particles were implicated in eliciting more pronounced pulmonary and cardiovascular effects following inhalation exposure (Seaton et al., 1995), researchers began to correlate impacts with exposure to ultrafine particles (Brown et al., 2002; Chalupa et al., 2004; Pekkanen et al., 2002; Wichmann and Peters, 2000). Although clear associations between ultrafine particle exposure and health impacts remained uncertain, this research was suggestive of a link between aerosol inhalation and health impacts that was mediated by particle size as well as chemistry, with smaller particles exhibiting a higher degree of potency. In this respect, epidemiological studies began to complement contemporary toxicology studies on inhalation exposure to fine particles.

These two streams of research began to coincide in the late 1990’s. But it was the formal advent of the field of nanotechnology toward the end of the 1990’s that galvanized action toward developing a more complete understanding of how material physicochemical characteristics impact on material hazard and how nanoscale materials might lead to previously unanticipated health impacts.

In the 1990’s, federal research agencies in the United States began looking to identify and nurture a new focus for science, engineering, and technology that would stimulate research funding and lead to economic growth. At the time, advances across the physical sciences were leading to breakthroughs in understanding of how material structure at the near-atomic scale influenced functionality and how this nanoscale structure might be intentionally manipulated. Recognizing the potential cross-disciplinary and cross-agency significance of these breakthroughs, an Interagency Working Group on Nanotechnology was established to promote the science and technology of understanding and manipulating matter at the nanometer scale (IWGN, 1999).

Although not fully realized until late in the 20th century (the first documented coining of the term “nanotechnology” is often credited to N. Taniguchi [Taniguchi, 1974]), the field of nanotechnology had its roots in 20th century advances in materials science and high-resolution imaging and analytical techniques. As techniques such as X-ray diffraction and transmission electron microscopy began to illuminate the structure of materials at the atomic scale—and how this structure influenced functionality—interest grew in improving materials through manipulating this structure. The fields of materials science and synthetic chemistry began to explore how small changes in structure at the atomic and molecular level could alter behavior at the macroscale. But it was perhaps the physicist Richard Feynman who first articulated a grander vision of nanoscale engineering. In a 1959 lecture at Caltech titled “There's plenty of room at the bottom,” Feynman speculated on the revolutionary advances that could be made if scientists and engineers developed increasingly sophisticated control over how substances were built up at the nanoscale (Feynman, 1960)—a level of control which at the time remained largely out of reach. Despite Feynman's lecture often being considered the foundation of modern nanotechnology, there is little evidence that it had much impact at the time (Toumey, 2008, 2010). However, the advent of Scanning Probe Microscopy in 1982 (Binnig et al., 1982), together with advances throughout the physical and biological sciences in imaging and understanding the nature of matter at the nanometer scale, began to open up the possibility of altering the functionality of a wide range of materials through nanoscale engineering.

Some of the more extreme and speculative possibilities of building materials and even devices molecule by molecule were captured in the popular book “Engines of Creation” by Eric Drexler, inspired by shrinking human-scale materials engineering down to the nanoscale (Drexler, 1986). Although many of the ideas put forward by Drexler were treated with caution and occasionally skepticism by the scientific community, there was a ground swell of excitement through the 1980’s and 1990’s over the possibilities that emerging techniques were opening up to systematically manipulating matter at the nanoscale, allowing nanoscale structure-mediated functionality to be exploited at the macroscale. This excitement was buoyed up by the discovery of carbon nanotubes (Iijima, 1991)—a new and functionally unique allotrope of carbon—and the demonstration of single-atom manipulation using scanning probe microscopy (Eigler and Schweizer, 1990). Working at this scale, new opportunities were arising for enhancing the structure of materials, for engineering materials tailored to exhibit specific physical, chemical, and biological behavior, for exploiting novel electron behavior in materials that begins to dominate at nanometer length scales, and for building increasingly sophisticated materials that could demonstrate multiple and context-specific functionality. The door was being opened to a new era of enhancing existing materials and products and creating innovative new ones by intentionally manipulating the composition and physical form of substances at the nanoscale.

Riding the wave of this cross-disciplinary “revolution” in science, engineering, and technology, President Clinton announced a new U.S. initiative to explore and exploit the science and technology of the nanoscale on January 21, 2000 (Clinton, 2000). In an address at Caltech on science and technology, he asked his audience to imagine “materials with 10 times the strength of steel and only a fraction of the weight; shrinking all the information at the Library of Congress into a device the size of a sugar cube; detecting cancerous tumors that are only a few cells in size,” and laid the foundation for the U.S. National Nanotechnology Initiative (NNI). Since then, the NNI has set the pace for national and international research and development in nanoscale science and engineering and has led the world in generating and using new knowledge in the field of nanotechnology.

As nanotechnology began to gain ground, it did not take long for concerns to be raised over the potential health and environmental implications of nanotechnology. In 2000, the cofounder of Sun Microsystems Bill Joy wrote an influential essay for Wired Magazine titled “Why the Future Doesn't Need Us” in which he raised concerns about the impacts of nanotechnology (Joy, 2000). This was followed by calls for a moratorium on research until more was known about the possible adverse impacts by one Civil Society group (ETC Group, 2003). More scientifically, sound concerns were raised by the reinsurance company Swiss Re in 2004 (Hett, 2004), and later that year, the UK Royal Society and Royal Academy of Engineering launched a highly influential report on the opportunities and uncertainties of nanotechnology (RS/RAE, 2004). At the center of the Royal Society and Royal Academy of Engineering report were concerns that engineered nanoscale materials with unique functionality may lead to unexpected exposure routes, may have access to unanticipated biological compartments, and may exhibit unconventional biological behavior associated with their size. In particular, concern was expressed over materials intentionally engineered to have nanoscale structure—nanomaterials—and particles and fibers with nanometer-scale dimensions—nanoparticles and nanofibers.

The Royal Society and Royal Academy of Engineering report marked a move toward a more integrated approach to the potential risks associated with nanotechnology. As global investment in nanotechnology research and development has grown (it has been estimated that global research and development investment in nanotechnologies exceeded $18 billion in 2008 and that the value of products utilizing these technologies in some way has been projected to exceed $3 trillion by 2015 [Lux Research, 2009]), so has interest in identifying, understanding, and addressing potential risks to human health and the environment (Chemical Industry Vision 2020 technology Partnership and SRC, 2005; ICON, 2008a; Luther, 2004; Maynard, 2006; Maynard et al., 2006; NNI, 2008; Oberdörster et al., 2005; PCAST, 2010; RCEP, 2008; SCENIHR, 2005, 2009). This interest has been stimulated by concerns that novel materials have the potential to lead to novel hazards and risks. But fueling it has been the research noted earlier on the role of particle size, physical form, and chemistry in mediating biological interactions and responses. With the advent of nanotechnology and the production of increasingly sophisticated engineered nanomaterials, research strands developing an understanding of the potential human health impacts of fine particles were thrust into the mainstream and became the basis of new thinking about how potential risks associated with new materials can be addressed.

As research began to focus on the potential hazards presented by engineered nanomaterials, the term “nanotoxicology” began to be used informally to describe this growing area of study. This was formalized in an editorial in Occupational and Environmental Medicine by Donaldson et al. (2004). Writing about the human health challenges presented by the emerging field of nanotechnology, Donaldson et al. noted that:

“NP [nanoparticles] have greater potential to travel through the organism than other materials or larger particles. The various interactions of NP with fluids, cells, and tissues need to be considered, starting at the portal of entry and then via a range of possible pathways towards target organs. The potential for significant biological response at each of these sites requires investigation. In addition, at the site of final retention in the target organ(s), NP may trigger mediators which then may activate inflammatory or immunological responses. Importantly NP may also enter the blood or the central nervous system, where they have the potential to directly affect cardiac and cerebral functions. We therefore propose that a new subcategory of toxicology—namely nanotoxicology—be defined to address gaps in knowledge and to specifically address the special problems likely to be caused by nanoparticles.”

The new field was consolidated in 2005 with a highly cited paper by Oberdörster et al. titled “Nanotoxicology: an emerging discipline evolving from studies of ultrafine particles” (Oberdorster et al., 2005), and the launch of the journal Nanotoxicology in 2007.

Since the early 2000’s, research into the potential impacts of nanomaterials and nanoparticles in particular has increased substantially. In the United States, the combined investment across federal agencies in research and development addressing environmental health and safety implications of nanotechnology was $34.8 million in 2005 (NSET, 2006). In 2011, this figure is estimated to rise to $116.9 million (NSET, 2010). Global publications addressing human health and environmental impacts of engineered nanomaterials have similarly increased. In 2005, there were an estimated 179 articles published on the potential environmental health and safety implications of engineered nanomaterials. By 2009, that number had risen to 791 publications (PCAST, 2010). Of these, the majority address the potential hazards of engineered nanomaterials. A search for publications with the key terms “nano*” and “toxic*” between 2000 and 2010 shows a rapidly increasing peer review literature in this area (Fig. 1)

FIG. 1.

Publications related to nanotoxicology, 2000–2010. Source: ISI Science Citation Index (Expanded). These data include research related to environmental and human health impacts, as well as toxicology-related research on nanoscale therapeutics, and thus provide an indicative rather than quantitative perspective on publications addressing nanomaterial toxicity in humans. “† ”Denotes data for 2010 were collected on September 19 and were pro rata'd for the full year to allow comparison with previous years. Actual 2010 data: “nano* AND toxic*”: 1364 publications; “nanotoxicology”: 64 publications.

Yet for all this activity, the field of nanotoxicology is suffering from something of an identity crisis. There is a strong sense that emerging, novel and complex materials that have been engineered at the nanoscale may exhibit unusual or unanticipated toxicity from a conventional perspective and that research is needed to understand and address how these designed materials might cause harm in ways that are not readily understood at present. This concern is supported by a growing body of research which indicates that some nanometer-scale materials do demonstrate biological behavior that is mediated by physical form as well as chemical composition (Donaldson et al., 2010; Nel et al., 2006; Oberdorster, 2010). Yet a clear identification and formulation of the problems being faced remain elusive. For example, what is meant by the “nanoscale” is far from clear, meaning that there is considerable ambiguity over which materials are embraced by “nanotoxicology.” Widely accepted definitions of nanotechnology refer to a size range of approximately 1–100 nm “where unique phenomena enable novel applications” (NSET, 2010). Yet these are largely definitions of convenience, not of science. And while the definitions defining the field of nanotechnology have been important in driving new science and technology innovation, it is not clear how they apply to a new material's propensity to cause harm in unexpected ways.

Within generally accepted definitions of nanotechnology, there is considerable ambiguity over the terms “uniqueness” and “novelty”—and how these attributes might lead to new materials that raise new health concerns. To a degree, “nanotoxicology” has been underpinned by an assumption that materials engineered to utilize unique properties associated with the nanoscale must, by definition, exhibit nanoscale-specific toxicity. Yet this assumption is far from secure. Indeed, a body of research has suggested that the toxicity of many nanomaterials is scalable—and thus predictable—from non-nanoscale materials (Oberdörster et al., 2007), questioning the uniqueness of the nanoscale. This does not of course negate the importance of studying nanomaterial toxicology—it simply brings into questions some of the blanket assumptions that direct this research. One of these assumptions is that the toxicity of nanomaterials is dominated by “quantum effects”—an assumption that is currently not supported in simple terms by the literature.

There is also uncertainty over the relationship between emerging nanoscale materials and established nanomaterials, including natural nanomaterials that have been present throughout human evolutionary history and anthropogenic nanomaterials (whether engineered or produced as a by-product) that have been part of human exposure for decades and even centuries. Although the argument is often made that engineered nanomaterials are unique by nature of their intentionally designed functionality and their precise physicochemical form, the boundaries between engineered nanoscale materials and other nanoscale materials in the real world can become blurred very rapidly. For example, humans have developed as a species in the presence of airborne carbonaceous nanoparticles from combustion, and our bodies have evolved to handle exposure to such materials. Since before the industrial revolution, people have been exposed to airborne metal and metal oxide nanoscale particles from hot processes (Maynard and Kuempel, 2005), and while these materials are rarely innocuous, we have an understanding of how they impact on human health. Even some forms of intentionally engineered nanomaterials have been used for many decades—the product Aerosil from Evonik (formally Degussa), for instance, is a fumed silica intentionally engineered to have a primary structure of the order of a few nanometers in scale. Aerosil has been used commercially for over 60 years.

This context does not detract from the emerging challenges presented by increasingly sophisticated new materials. But it does demand that careful thought is given to the toxicity of these materials and whether they are genuinely an unknown quantity or whether we have a body of evidence and understanding from which to address them. And it does require a distinction to be made between the language and terminology that drives a new field of technology innovation such as nanotechnology and that which drives research into understanding potential health impacts. History suggests that not every new technology leads to new hazards and not every new hazard is associated with a new technology.

Nevertheless, there is an array of increasingly sophisticated materials that are emerging from advances in science, technology, and engineering that do demand careful consideration of the new risks they might pose. In this respect, a differential approach to toxicology studies is required—one which helps identify where emerging materials and products deviate from established ones in their potential to cause harm and focuses research on narrowing the resulting knowledge gap. Undoubtedly, materials intentionally designed and engineered to behave in specific ways because of their fine structure are at the forefront of the new challenges being faced in toxicology. These materials increasingly demonstrate biological behavior that results from a synergistic interaction between chemical composition and physical form. But whether these new challenges can be confined to a narrow size scale implied by “nanotoxicology” is debatable. Rather, we would argue that a broader perspective is needed on the challenges presented by novel and functional materials that capture the idea of “sophisticated materials.” These are substances that arise at the intersection of scientific disciplines and technology platforms and demonstrate novel and even time and context-dependent functionality based on their engineered and increasingly complex physicochemical structure. Although many of these materials will depend on nanoscale engineering, decoupling the materials from the underlying technology—or technologies—is helpful in formulating science-based questions regarding their toxicity. In this respect, the toxicology challenge presented by sophisticated materials is to understand and address the hazards presented by materials that have the ability to enter the body, interact with it, and elicit an adverse response in ways that are not adequately understood through a conventional and chemical composition–dominated perspective on toxicology.

In this review, we present a brief overview of the current state of the science on the toxicology of nanoscale materials and focus on three areas of emerging toxicology-based challenges presented by sophisticated materials: identifying relevant materials for study, physicochemical characterization, and biointeractions. Given the rapidly increasing breadth of research on the potential hazards and risks presented by engineered nanomaterials, a comprehensive evaluation of the field is beyond the scope of this review. It is also somewhat redundant, given the large number of excellent previously published reviews and analyses (Aitken et al., 2009; Balbus et al., 2007; ICON, 2008b; Maynard and Kuempel, 2005; Maynard et al., 2006; Oberdorster, 2010; Oberdorster et al., 2005, 2007; SCENIHR, 2005, 2009; Warheit et al., 2007). Rather, here we consider aspects of nanoscale materials that set them apart from more conventional materials and build on these to explore the emerging challenges of understanding the toxicology of sophisticated materials.

THE TOXICOLOGY OF NANOSCALE MATERIALS

In 2005, the European Commission Scientific Committee on Emerging and Newly Identified Health Risks (SCENIHR) published a comprehensive assessment of the state of the science regarding potential risks associated with “engineered and adventitious products of nanotechnologies” (SCENIHR, 2005). It was one of the first in a long series of assessments and reviews of the toxicology of nanoscale materials that have helped identify emerging issues surrounding the potential health impacts of these materials, and although the state of the science has moved on since its publication, the overarching issues identified by the committee remain contemporary.

The SCENIHR committee was tasked with addressing three questions: Are existing methodologies appropriate to assess potential and plausible risks associated with different kinds of nanotechnologies and processes associated with nanosized materials as well as the engineered and adventitious products of nanotechnologies? If existing methodologies are not appropriate to assess the hypothetical and potential risks associated with certain kinds of nanotechnologies and their engineered and adventitious products, how should existing methodologies be adapted and/or completed? And in general terms, what are the major gaps in knowledge necessary to underpin risk assessment in the areas of concern?

In common with most other reviews addressing the toxicity of nanomaterials, SCENIHR focused on materials that are physically able to enter the body via inhalation, ingestion, and potentially dermal penetration, leading to an emphasis on the particulate form of nanomaterials—and nanometer-scale particles (nanoparticles) in particular. In reviewing the literature, the committee identified three nanoscale mediators of toxicity: particle size, shape, and chemical composition. Drawing on evidence of material toxicity that was influenced by physical form as well as chemical composition, SCENIHR explored how these three mediators potentially affect bioavailability and biointeractions and influence exposure and dose. Specific mechanisms of toxicity highlighted included epithelial tissue injury, inflammation, oxidative stress, and allergy. Concluding that “there is insufficient data available to identify any generic rules governing the likely toxicology and ecotoxicology of nanoparticles in general,” the committee identified a number of major knowledge gaps that prevented a complete risk assessment of engineered nanomaterials. These included understanding the mechanisms and kinetics governing nanomaterial release, quantifying the range of potential exposures, developing an understanding of the extent to which data from non-nanosized materials can be extrapolated to the nanoscale, generating toxicokinetic data associated with various portals of entry to the body, and addressing worker health.

Although the SCENIHR report was published nearly 6 years ago, it outlined issues associated with synergistic interaction between the chemical composition and physical form of nanoscale materials and their biological interactions that continue to be relevant. Although the state of the science has moved on since the report's publication, the key themes that the committee laid out remain central to understanding and addressing the toxicology of nanoscale materials.

These themes are reflected and expanded on in one of the more recent reviews of the field by Oberdörster (Oberdorster, 2010). Although there is a growing literature on the toxicology of nanoscale materials and many reviews of the potential risks presented by such materials (Aitken et al., 2009; Balbus et al., 2007; Donaldson and Poland, 2009; Maynard, 2006, 2007a, 2007b; Maynard and Kuempel, 2005; Maynard et al., 2006; Nel et al., 2006; Oberdorster et al., 2005), much of this is captured in Oberdörster's review.

Oberdörster considers the toxicology of nanoparticles (as a special but biologically important case of nanomaterials) in terms of their physicochemical characteristics, their biokinetics, and their effects. Specifically, he focuses on nanoparticles that are likely to be biopersistent and therefore show prolonged behavior that is governed by their physicochemistry. Relatively transient nanoparticle such as nanoscale micelles and liposomes are not addressed—whereas the temporal physical form of these and similar “soft” materials may influence their toxicity, it remains unclear the extent to which their impact is dominated by chemistry or form.

Comparing particles smaller than 100 nm in diameter to those > 500 nm in diameter, Oberdörster identifies 22 aspects that are potentially important to influencing size-related biological impact (Table 1). In doing so, he begins to develop a framework for a differential toxicology approach to nanomaterials, where the toxicology of nanoscale materials is understood in the context of chemically similar but physically different materials. Importantly, this approach acknowledges the fuzzy transition between large and small particles that is not always governed by well-defined size boundaries and abrupt changes in behavior.

TABLE 1.

Comparing the Characteristics, Biokinetics, and Effects of Inhaled Nanoparticles versus Larger Particles (Oberdorster, 2010)

	Nanoparticles (< 100 nm)	Larger particles (> 500 nm)
General characteristics
Ratio: Particle number/mass or surface area/mass	High	Low
Agglomeration/aggregation in air and/or liquids	Likely (dependant on medium)	Less likely
Deposition in respiratory tract	Diffusion dominates	Sedimentation, impaction and interception dominate
Protein/lipid adsorption in vitro	Yes; important for biokinetics	Less important
Translocation to secondary target organs
Clearance	Yes	Generally not
Mucociliary	Probably yes	Efficient
Alveolar macrophages	Poor	Efficient
Epithelial cells	Yes	Mainly under overload
Lymphatic circulation	Yes	Under overload
Blood circulation	Yes	Under overload
Sensory neurons (uptake and transport)	Yes	No
Protein/lipid adsorption in vivo	Yes	Some
Cell entry/uptake	Yes (caveolae, clathrin, lipid rafts, diffusion)	Primarily phagocytic cells
Mitochondria	Yes	No
Nucleus	Yes (< 40 nm)	No
Direct effects (chemistry and dose dependent)
At secondary target organs	Yes	No
At portal of entry (respiratory tract)	Yes	Yes
Inflammation	Yes	Yes
Oxidative stress	Yes	Yes
Activation of signaling pathways	Yes	Yes
Primary genotoxicity	Some	No
Carcinogenicity	Yes	Yes

Within this framework, Oberdörster highlights three areas which are significant in understanding nanomaterial toxicity compared with that of macroscale materials and/or constituent chemicals: dose, biokinetics, and the significance of physicochemical properties. Although these are not the only issues of significance in addressing nanomaterials, they provide a useful framework for summarizing the current state of the science.

Dose

Over the past 20 years, questions surrounding dose, including how it is characterized and quantified, have been central to addressing the toxicity of nanomaterials. As was highlighted earlier, evidence has emerged that, for some materials, the use of mass concentration alone as a dose metric can obscure associations between the material and biological behavior. If response is mediated by particle number concentration, the disparity between what is measured and what leads to an affect is potentially large if mass is the dose metric used as the number of particles in a given mass of material increases inversely with diameter cubed. For example, 1 mg of 10 μm diameter spherical carbonaceous particles would consist of approximately 10¹² particles; the same mass of 10-nm diameter particles would consist of approximately 10²¹ particles. A smaller but still potentially significant disparity exists if mass is used as a dose metric where surface area mediates response. For a given mass of material, surface area varies inversely with particle diameter (assuming spherical particles). So whereas 1 mg of 10-μm diameter spherical carbonaceous particles has a surface area of approximately 270 m², the same mass of 10-nm diameter particles has a surface area of 270,000 m². As a result, although Paracelsus observation that “the dose makes the poison” still holds in contemporary toxicology, there is considerable uncertainty over what is meant by dose when it comes to nanomaterials.

A number of studies have suggested that particle surface area is a relevant metric for small, insoluble inhaled particles (Maynard and Kuempel, 2005; Oberdörster, 2000). Yet it is by no means clear whether this is a general rule for a wide range of materials and exposure routes. Even with well-studied materials such as TiO₂, there is research, suggesting that surface area alone may not provide a good indicator of response (Warheit et al., 2006). It is also possible that conventional metrics of mass concentration and chemical composition may be used as surrogate measures of dose, even when effects are not driven by the measured quantity per se (Maynard and Aitken, 2007). For instance, in a highly monodisperse suspension of nanoparticles, dose characterized by mass, surface area, or particle number are highly correlated and probably interchangeable.

In addressing how dose is most appropriately characterized, there remains limited understanding of the underlying mechanisms of interaction and impact. For instance, where surface area correlates well with response, there is uncertainty whether (in specific cases) this is governed by dissolution, surface reactivity, or other mechanisms. A greater understanding is needed of these mechanisms before empirical findings on dose-response for engineered nanomaterials can be placed on a more systematic and mechanistic footing. This will become increasingly important as more sophisticated materials are engineered with complex and multifunctional components at the material-biological interface.

An issue related to dose metrics raised by Oberdörster is that of dosimetry. Oberdörster argues that an increasingly sophisticated understanding of dosimetry is needed—one that not only recognizes different mediators of response but also one that is related to real-world exposures and is responsive to localized dose within the body. There have been a number of instances where in vitro studies have been published demonstrating a response to nanoparticles, but at doses that far exceed those reproducible in vivo (Oberdorster, 2010)—resulting in headline-catching data that is difficult to interpret and near-impossible to apply to human exposures. Similarly, there have been in vivo studies that elicit responses at extremely high doses but are again difficult to relate to real-world conditions precisely because of this. As Oberdörster notes, these studies are valuable in exploring proofs of principle but are limited in terms of their ability to develop a clear and predictive understanding of nanomaterial toxicity. This becomes all the more difficult if the dose metric of relevance is not the one that is measured, leading to the possibility of unintended high dosing in studies.

Questions surrounding dosimetry also relate to localized and temporal dose. If nonlinear associations exist between dose and response, significant spatial and time variations in dose within an animal model or cell culture have the potential to confound studies. For example, the administration of an aerosol as a high concentration bolus in inhalation studies has the potential to influence response and lead to data misinterpretation. Oberdörster cites a study where a total dose of 7.5 mg of nanoscale TiO₂ was instilled intranasally in mice and resulted in significant oxidative stress and inflammation in the brain (Wang et al., 2008). The study was subsequently highlighted in the media, where it was misrepresented as an inhalation study showing that nanoparticles can damage brain cells (Benninghoff and Hessler, 2008). As Oberdörster points out, the dose in this case was the equivalent of intranasally instilling ∼17.5 g of the material into a human subject.

In addition, localized dose “hot spots” often drive response following aerosol inhalation. Particles preferentially deposit at bifurcations in the airways—large particles through inertial deposition and small particles through diffusion within the stagnation zone that develops at bifurcations. These localized doses—which can be a hundredfold higher than the mean dose for larger particles—are frequently used to justify high dose in vitro assays. Yet the local dose enhancement for nanoparticles is somewhat different—ranging from around a 5-fold to a 60-fold increase in dose (Balásházy et al., 2003). Misrepresenting these dose “hot spots” as they relate to particle size has the potential to confound the extension of in vitro studies to in vivo exposures.

The question of dose also becomes important when comparing studies and when developing predictive models of nanoparticle toxicity. This is particularly significant when comparing in vitro and in vivo studies, where physicochemical parameters make simple comparisons difficult. Rushton et al. (2010) have proposed a novel approach where studies are compared using the steepest part of the dose-response curve. Using this approach, Rushton et al. have reported good predictive power between in vitro cell-free studies and in vivo studies looking at inflammatory response. Building on this work, the authors have looked at using the maximum rate of response as a function of dose (the steepest part of the dose-response curve) as an approach to categorize nanomaterial hazard based on reactivity per unit surface area (Rushton et al., 2010).

As a final reflection dose, there is increasing evidence that particulate dose may need to be rethought in in vitro studies as well as in vivo studies. Teeguarden et al. (2007) have identified discrepancies between the amount of a material introduced to in vitro cell cultures—nominally considered to be the dose—and the amount of material cells are able to interact with. As particles form a dynamic concentration gradient within the suspension medium, there are indications that over short time periods actual doses of material experienced by cells may be orders of magnitude lower than assumed, suggesting that further work is needed in characterizing particle doses in vitro.

Physicochemical Properties

The biological nature of nanoscale materials is intimately associated with their physical form and chemical composition, leading to toxicologic responses that are associated with a wide range of physicochemical parameters and that are affected by dynamic changes in materials. Understanding the association between physicochemical properties, biological interactions, and hazard is a significant challenge as it requires new approaches to think about how physical form—which may vary with time and between batches of material—can modulate biological response beyond what is anticipated from chemical composition alone. In 2005, Oberdörster et al. proposed 17 physicochemical material characteristics that potentially affect nanomaterial toxicity and which ideally need to be characterized in studies (Oberdorster et al., 2005). Recognizing the dynamic nature of these materials, characterization in situ was recommended where possible, as well as characterization of the as-supplied material and the as-administered material. This list of parameters formed the basis of a reduced list developed at a workshop held in Washington, DC, in 2008, and made public through the Minimum Information for Nanomaterial Characterization (MINChar) Initiative (2008). Similar lists have been proposed in the literature (Card and Magnuson, 2009; Warheit, 2008).

Particle aggregation and agglomeration present particular challenges in toxicology studies. The process of particles joining together to form weak bonds (agglomeration) or strong bonds (aggregation) changes profoundly the size, dynamics, and properties of the resultant clusters. In air, changes in particle size through agglomeration influence transport, deposition, whether the material can be inhaled, where it deposits within the respiratory tract, whether it can translocate from the lungs to other parts of the body, and how it is cleared from the body. Likewise, agglomeration and aggregation in liquids affects how a material is transported, where it goes, and how it interacts with its environment. Agglomeration/aggregation (or even de-agglomeration) between material release, exposure, and transport within the body (or preparation, administration, and transport in toxicology studies) may lead to significant changes in hazard potential. For instance, where transport between organs, across cell barriers, and along neuronal pathways is mediated by particle size, an understanding of agglomeration/aggregation state is essential to understanding potential impact (Oberdorster, 2010). The rate at which particles will aggregate or agglomerate is dependent on concentration and size—the smaller the particles and the higher the concentration, the greater the aggregation/agglomeration rate (Hinds, 1999).

Internal particle structure has also been shown to influence toxicity. Jiang et al. (2008) and Sayes et al. (2006) have shown for instance that the crystal structure of TiO₂ nanoparticles can have a significant impact on particle toxicity. In both studies, anatase TiO₂ was found to be more potent than the rutile form of the material. Mixtures of anatase and rutile TiO₂ had an intermediate potency. Using a cell-free assay designed to probe a material's capacity to generate reactive oxygen species (ROS), Jiang et al. also indicated a significant dependence between particle size and capacity to generate ROS, with a clear transition in behavior with anatase nanoparticles between ∼10 and 40 nm. What was particularly interesting in this study was that the smallest particles demonstrated a reduced capacity to generate ROS. However, as the surface structure of materials can change markedly at very small sizes (Jefferson, 2000), it is unclear whether this transition was size mediated or surface chemistry mediated. The authors speculated that the findings might be associated with the density of defects on the surface of the particles, suggesting another physicochemical parameter of potential interest in understanding the toxicity of nanomaterials.

As well as particle size, particle shape has also been indicated as a key parameter in determining biological impact. In particular, the fiber-like morphology of some carbon nanotubes has prompted concerns over possible asbestos-like behavior following inhalation, including the potential development of mesothelioma (Coles, 1992; Maynard et al., 2006; RS/RAE, 2004). Takagi et al. induced mesothelioma and reduced mortality in p53+/− mice through ip injection of multiwalled carbon nanotubes (Takagi et al., 2008), although this study was subsequently criticized for the use of extremely high doses and poor material characterization (Ichihara et al., 2008). To confirm the possibility of mesothelioma resulting from exposure to carbon nanotubes, Poland et al. exposed the mesothelial lining of the peritoneal cavity of mice to long multiwalled carbon nanotubes via ip injection and concluded that the early pathological effects were characteristic of asbestos-like events in producing inflammation (Poland et al., 2008). Subsequently, Ryman-Rasmussen et al. (2009) subjected mice to a single inhalation exposure of multiwalled carbon nanotubes and reported that, at a subsequent post-exposure period, the nanotubes translocated from airspace to sites outside the respiratory tract and embedded in the subpleural wall and within subpleural macrophages. This finding served to provide an indirect confirmation of the possibility of inhaled multiwalled carbon nanotubes producing effects both inside and outside the respiratory tract—similar to asbestos fibers. It is interesting to note, however, that two 90-day inhalation studies with multiwall carbon nanotubes conducted in rats, reported by Ma-Hock et al. (2009) (Nanocyl multiwalled carbon nanotubes) and Pauluhn (2010) (Baytubes multiwalled carbon nanotubes), failed to find pathological effects outside the respiratory tract. Either there is a difference among species, the pleural effect is not particularly pronounced or a greater focus needs to be implemented to investigate the potential and relevance of this pleural effect (Warheit, 2009). To add further complexity to the biological actions of nanotubes, Kagan et al. (2010a) recently reported that carbon nanotubes may be biodegraded via a neutrophil myeloperoxidase mechanism under conditions of inflammation, although it remains unclear how relevant the results of this in vitro study are to conditions in vivo.

The question over carbon nanotube toxicity is dominated by the physicochemical nature of the material. Carbon nanotubes are not a homogeneous material category but rather represent an extremely wide array of material chemistries and morphologies, determined by the number of concentric graphene walls constituting the nanotubes, their chirality, their diameter, their length, the density of surface defects, surface functionalization, the presence of trace elements and other contaminants, nanotube straightness, the degree of nanotube entanglement, and so on. Poland et al. demonstrated the potential of one subset of this material—long, straight multiwalled carbon nanotubes—to show fiber-like behavior in a biological environment (Donaldson et al., 2010). However, many forms of the material are too short, too long, or too tangled to demonstrate similar behavior. Nevertheless, these non-fiber–like forms of carbon nanotubes may present their own distinct hazards (Shvedova et al., 2003, 2008). Given evidence that the morphology of carbon nanotube material released into the air during handling can vary markedly from batch to batch, the challenges of relating relevant characteristics to hazard are complex (Maynard et al., 2007). This is a material that cannot be adequately characterized by chemistry alone, or as a simple fiber, in determining its potential toxicity. Rather, it epitomizes the need for a detailed and sophisticated understanding of nanomaterial physicochemical characteristic in understanding potential hazard.

Biokinetics

Unlike free or loosely bound molecules, the transport, accumulation, transformation, and clearance of nanomaterials in the body is intimately associated with physical form as well as chemical composition. Understanding the biokinetics of nanomaterials provides information on internal doses to secondary organs and is essential to designing and interpreting in vitro studies. Oberdörster cites the well-documented tendency of nanoparticles to translocate from primary deposition sites to secondary organs (Oberdorster, 2010) but cautions that uninformed interpretation of these data can lead to misunderstanding of potential risk. Inhalation studies using 15 and 80 nm iridium nanoparticles have demonstrated the translocation of inhaled particles to extrapulmonary organs. However, translocation rates were the order of ∼1–2%, with the rate decreasing rapidly at larger particle sizes (Kreyling et al., 2002, 2009). Nevertheless, there is mounting evidence that changes in physical and chemical nature at the nanoscale can have a significant impact on biodistribution. For example, Semmler-Behnke et al. (2008) have demonstrated a marked difference in biodistribution of 1.4-nm diameter Au₅₅ clusters and 18-nm diameter gold particles administered to rats via injection and intratracheal instillation; 24 h following iv injection, 18-nm diameter gold particles were cleared from the blood and predominantly accumulated in the liver and spleen; 0.5% of the injected dose was excreted via that hepatobiliary system, but renal excretion was extremely low. In comparison, the 1.4-nm diameter gold clusters were excreted by the kidneys as well as by the hepatobiliary system.

Of particular concern in recent years has been the nature of interactions between nanoparticles and the central nervous system (Yang et al., 2010). There is evidence that inhaled nanoparticles can translocate to the central nervous system via olfactory neurons following nasal deposition (Oberdörster et al., 2004) and induce significant inflammation-related effects (Elder et al., 2006). This appears to be a particle size and chemistry transport route that is unique to nanometer-scale particles and raises the possibility of previously unidentified organ-specific doses and responses. Although data remain inconclusive, Oberdörster hypothesizes that differential protein adsorption on nanoparticles will affect their uptake and transport within the central nervous system (Oberdorster, 2010). Preliminary data generated using Apolipoprotein E–coated gold nanoparticles are consistent with increased nanoparticle translocation to the central nervous system in rats following iv administration. However, in this study, less than 0.01% of injected particles were translocated, leaving the authors to conclude that further confirmatory studies are needed (Oberdorster, 2010).

Nanoparticle translocation to the central nervous system is indicative of research on a number of fronts looking at nanoparticle movement across tight barriers. For a number for years, there has been concern over the ability of blood-borne nanoparticles to cross the placental barrier (Saunders, 2009). Recently, Bhabra et al. (2009) have indicated – using an in vitro model – that blood-borne nanoparticles may be able to exert an influence across the placental barrier without physically crossing it. Using an in vitro system designed to investigate cellular barriers, Bhaba et al. showed that high concentrations of Cobalt-Chromium alloy nanoparticles on one side of a tightly meshed layer of cells can cause measurable DNA damage to cells on the other side. However, it remains uncertain how relevant these data are to in vivo exposures.

The skin represents another tight barrier that has received a high level of attention in recent years, as concerns over the ingress of mineral nanoparticle such as TiO₂ and ZnO from sunscreens and cosmetics into the body have been raised. Early research suggested that the potential exists for sub-micrometer diameter particles to penetrate across the dermal barrier under some circumstances, depending on their size and chemistry (Ryman-Rasmussen et al., 2006; Tinkle et al., 2003). However, the majority of studies to date suggest that under most conditions healthy skin is an effective barrier to nanometer-scale particles entering the body and causing adverse effects (Choksi et al., 2010; Newman et al., 2009; Nohynek et al., 2008, 2010; Osmond and McCall, 2010; Stern and McNeil, 2008). However, there remain some uncertainties surrounding the tightness of the skin as a barrier against nanoparticles under varying conditions, and the impacts and clearance of particles that may cross the barrier. Recently, Gulson et al. (2010) exposed human volunteers to sunscreens formulated with ZnO particles tagged with the stable isotope ⁶⁸Zn. Traces of ⁶⁸Zn were found in blood and urine samples of volunteers exposed to nanometer-scale and non-nanoscale particles, providing evidence of Zn transport into the body. However ⁶⁸Zn levels were orders of magnitude below normal blood-borne Zn concentrations. It also remains uncertain whether these findings were associated with particle translocation, or particle dissolution and subsequent ion transport.

As has been alluded to, a possible confounding factor in understanding the biokinetics of nanoparticles is their differential interactions with proteins within biological environments. Nanoparticles within a biological environment rapidly acquire a coating or “corona” of protein molecules and there is increasing evidence that this dynamic coating mediates the transport of, and first order interactions with, nanoparticles within the body (Cedervall et al., 2007a, 2007b; Ehrenberg et al., 2009). Furthermore, there is evidence that the corona – and thus particle biokinetics – is influenced by particle size and chemistry (Lundqvist et al., 2008). This relatively new area of research suggests that interactions between nanoparticles and biological systems may be more complex and dynamic than previously thought, requiring a more holistic understanding of how biokinetics are influenced by particle physicochemistry and their local environments over time.

EMERGING CHALLENGES

Although we have touched on just some of the more prominent developments in the science of nanoscale materials toxicology, it is clear that as understanding of how these materials interact with biological systems increases, new questions are being raised as to how to understand and quantify the toxicity of increasingly sophisticated materials in the context of identifying, assessing, and managing risks. It is also becoming clear that, although new questions are being prompted by the development and commercial use of engineered nanomaterials, the challenges being faced by toxicology are not solely confined to materials or particles with physical structure in the range of 1–100 nm. Rather, the emergence of new nanomaterials is highlighting the importance of material physicochemistry in mediating biological interactions that result in toxicity. Research to date suggests that synergism between particle chemistry and physical form becomes increasingly important as the features and dimensions of materials entering the body become increasingly small. But beyond this, there are few indicators of generalized sharp size-specific transitions in behavior. Aufann et al. (2009). have attempted to define a particle size region where size-specific biological behavior unique to nanoparticles might occur. Reviewing the literature, they concluded that particles smaller than ∼30 nm in diameter are more likely to demonstrate dramatic changes in behavior with size. However, particle sizes at which abrupt changes in behavior occur are clearly material dependent—as was shown by Semmler-Behnke et al. in contrasting the biokinetics of 1.4 and 18 nm diameter gold nanoparticles (Semmler-Behnke et al., 2008). And there is little reason not to suppose that some materials may exhibit abrupt changes in behavior above 100 nm.

Concerns over the possible novel toxicity of nanomaterials are frequently driven by abrupt size-specific changes in functionality that are governed by size-constrained electron behavior—often referred to as “quantum effects.” Yet very few studies have shown a clear association between nanoscale phenomena such as quantum confinement or surface plasmon resonances and toxicity. Instead, studies have tended to highlight the importance of decreasing particle size and increasing specific surface area for specific particle chemistries in altering biological behavior. In many cases, these are scalable effects—small particles show greater or less tendency to behave in a certain way compared with large particles, but their behavior is predictable from larger particles. This is the case for most studies correlating toxicity with surface area. In other cases, nonscalable effects are seen, such as with size-specific translocation. Yet even here, it is unclear whether the unusual biological behavior observed is related to the functionality these materials are designed to exhibit or simply a function of small size.

Yet the field of toxicology is undoubtedly facing a new and growing challenge: How to understand and address the hazard of intentionally engineered materials where physical form and chemical composition interact synergistically to determine biological behavior. As collaborations across diverse fields of research lead to increasingly sophisticated new materials—many of which will be engineered with nanoscale features—this challenge will only grow in magnitude. Up to now, research has been driven by small particles of conventional materials such as TiO₂, ZnO, and Ag and the occasional new material such as carbon nanotubes. But as the science and technology of new materials becomes increasingly sophisticated, toxicologists will be faced with complex multicomponent materials, hybrid materials that blur the boundaries of biological and nonbiological components and active materials that are designed to change behavior according to their environment or a received set of signals (Subramanian et al., 2010). These new sophisticated materials will require a new toxicology that recognizes the significance of physicochemistry and dynamic (and possibly remote activated) behavior, that is cognizant of but not constrained by the importance of the nanoscale, and that is focused on the potential biological impacts of the materials rather than their commercially relevant functionality.

Within this context, we highlight three emerging challenges to addressing the toxicology of sophisticated materials: identifying materials that have the potential to exhibit novel and significant toxicity, characterizing materials appropriately, and biointeractions.

Identifying Relevant Materials

Effective problem formulation is a cornerstone of contemporary risk assessment and, by association, toxicology (National Academy of Science, 2008). Nevertheless, formulating the environmental health and safety impact “problems” posed by sophisticated materials is not trivial. A key question is how to delineate between the materials and products that are of concern and those that are not. In regard to engineered nanomaterials, the conventional approach has been to use established definitions of nanotechnology and engineered nanomaterials. These debates typically focus on material functionality within a narrow size range and are designed primarily to stimulate research and innovation leading to economically and socially beneficial new products (NSET, 2010). However, these simple function-oriented definitions do not always lend themselves to supporting well-defined problem statements that frame relevant toxicology research on engineered nanomaterials. For example, they do not allow easy differentiation between functionally unique materials and products that do not present new toxicology challenges and functionally mundane materials and products that do present new hazards. An example of the former might be cadmium selenide-based quantum dots, where functionality is associated with size-dependent quantum confinement, but hazard is more likely associated with the composition of the quantum dots. And an example of the latter might be the use of nanoscale particles in a product simply on the grounds of convenience or economy, but where particle size leads to new exposures, doses, and hazards. This disconnect between definitions driving research and innovation and hazard-based problem formulation is likely to become increasingly important in the face of increasingly sophisticated materials.

An alternative approach to addressing potential hazards presented by sophisticated materials is to use principles that guide scientifically grounded problem formulation. Three principles that go some way to support science-based and socially relevant problem formulation address emergent risk, plausibility, and impact.

Emergent Risk

The idea of emergent risk reflects the likelihood of a new material causing harm in a manner that is not apparent, assessable, or manageable based on current approaches to risk assessment and management. Examples of emergent risk include the ability of small particles to penetrate to normally inaccessible places, the inapplicability of established toxicology assays to some materials, scalable behavior that is not addressed by conventional approaches to assessing hazard, and the possibility of abrupt scale-dependent changes in material interactions within biological systems. This understanding of “emergence” is dependent on the potential of a material to cause harm in unanticipated or poorly understood ways, rather than its physical structure or properties per se. As such, it is not bound by rigid definitions such as those used to define nanotechnology or nanomaterials. Rather, it enables sophisticated materials that potentially present emergent and unanticipated risks to human health and the environment to be distinguished from those that probably do not.

Many of the engineered nanomaterials that have raised concerns in recent years have shown potential to lead to emergent risks and thus would be classified as requiring further investigation under this principle. But the principle also embraces more complex nanomaterials that are either in the early stages of development, or have yet to be developed, including active nanomaterials and self-assembling materials.

Plausibility

Plausibility captures—in qualitative terms—the science-informed likelihood of a new material or product presenting a risk to humans. It is based on the possible hazard of a material and potential for exposure or release to occur. But it also addresses the likelihood of a technology being developed and commercialized, and it leading to emergent risks. For example, the “gray goo” of self-replicating nanobots envisaged by some (Joy, 2000) might legitimately be considered an emergent risk but is clearly not a plausible risk. In this way, plausibility acts as a crude but effective filter to distinguish between speculative risks—which are legion—and credible risks—which are not.

Impact

Impact is an indicator of the extent to which a poorly managed sophisticated material might cause harm or the possible reduction in harm resulting from new research into identifying, assessing, and managing emergent risks. It helps provide a qualitative reality check to guard against extensive research efforts that are unlikely to have a significant impact on protecting human health, while ensuring that research having the potential to make a significant difference is identified and supported.

Together, these three principles provide a basis for developing informed and relevant approaches to problem formulation when faced with evaluating the hazards associated with emerging sophisticated materials. They are tools that allow new materials which raise safety concerns to be differentiated from those that, while they may be novel from an applications perspective, do not present undetected, unanticipated, or enhanced risks. The principles are technology independent and therefore can be used to guide research independently of the sophistication of the materials being produced or shifts in terminology and emphasis underlying technology innovation.

Applying the principles to increasingly sophisticated materials that are being envisaged, a number of groups of materials begin to emerge that may require further study:

Materials demonstrating abrupt scale-specific changes in biological or environmental behavior.

Materials that undergo rapid size-dependent changes in physical and chemical properties, which in turn affect biological behavior, may present a hazard that is not predictable from larger scale materials of the same composition. In this case, size and form at the nanoscale may increase or decrease hazard in a way that is currently not well understood.

Materials capable of penetrating to normally inaccessible places.

Materials that, by their size, shape, and/or surface chemistry, are able to persist in or penetrate to places in the body that are not anticipated based on current understanding may present emergent risks. Where there is a credible possibility of accumulation of, exposure to, or organ/system-specific dose associated with a material that is not expected from how the dissolved material or larger particles of the material behave, a plausible and emergent risk is possible.

Active materials.

Materials that undergo a change in their biological behavior in response to their local environment or a received signal (Subramanian et al., 2010), potentially present dynamic risks that are currently not well understood.

Self-assembling materials.

Materials designed to assemble into new structures in the body once released raise issues that may not be captured well within current approaches to hazard assessment.

Materials exhibiting scalable hazard that is not captured by conventional hazard assessments.

Where hazard scales according to parameters other than those normally associated with an assessment, emergent risks may arise as dose-response relationships are inappropriately quantified. For instance, if the hazard presented by an inhaled material scales with the surface area of the material and the dose-response relationship is evaluated in terms of mass concentration, the hazard will remain ill quantified.

In each of these examples (they are not exclusive), new research is needed if emergent and plausible risks associated with new sophisticated materials are to be identified, characterized, assessed, and managed.

Physicochemical Characterization

Relevant physicochemical characterization is essential to interpreting data from toxicity studies on sophisticated materials if generated data are to be useful. The early research by Oberdörster et al. on inhalation exposure to TiO₂ particles using rats showed that chemistry alone could not explain differences in dose-response relationships for two distinct sizes of particles with the same composition (Oberdörster, 2000); it was only when the physical structure of the two materials was included in the assessment that the data were reconcilable—and a single dose response relationship relative to material surface area emerged. But relevant physicochemical characterization is also necessary if different studies are to be reproduced and compared. Without it, vital information is lacking that can prevent a robust picture of material toxicity from emerging. In the case of the Oberdörster study, material surface area was measured, but not discrete particle size or aggregation state. As a result, it was initially difficult to evaluate or validate whether the effects observed were simply because of an elevated material surface area or were associated with the presence of discrete nanometer-scale particles. A clearer case of data confounding through a lack of physicochemical characterization can be found in studies on carbon nanotubes. Despite the enormous variation in physical and chemical properties among carbon nanotubes from different sources (or even the same source at different times), early toxicology studies were remarkably vague on the precise nature of the materials being studied, leading to irreproducible, conflicting, and ultimately uninterpretable data (Lam et al., 2006).

The relevance of physicochemical characterization in understanding and assessing material toxicity has received considerable attention in recent years. A workshop in 2004, organized by the National Institute of Environmental Health Sciences and the University of Florida placed a strong emphasis on the need for highly detailed materials characterization for instance (Moudgil, 2004). These recommendations—driven in part by materials scientists—were considered at the time to be beyond the scope of many toxicologists. In 2005, an influential review led by Oberdörster proposed a reduced—but still extensive—set of physicochemical parameters that should be included in nanomaterial toxicology studies (Oberdorster et al., 2005). The recommendations listed 17 parameters as either essential or desirable in studies. Importantly, they also considered characterization at three distinct points, recognizing the dynamic nature of physicochemical characteristics: in the bulk material, as-prepared for administration, and in situ. Of these, the importance of characterizing materials in situ was stressed as materials are capable of undergoing significant alterations in properties once they are introduced to a biological environment. However, it was also recognized that few technologies currently exist that enable detailed materials characterization within in vitro and in vivo test systems.

Although the Oberdörster recommendations were less challenging than those from Florida, they were still seen as presenting near-insurmountable barriers to toxicologists. As a result, discussions within the community continued to focus on a minimum characterization set that would be both feasible and readily adoptable. A 2008 workshop was held at the Woodrow Wilson International Center for Scholars in Washington, DC, was influential in mapping out such a minimum characterization set (Table 2) (MINChar Initiative, 2008). The list was one of the first pragmatic sets of physicochemical characterization requirements and reflects similar thinking published elsewhere (Boverhof and David, 2010; Card and Magnuson, 2009; Warheit, 2008).

TABLE 2.

Proposed Minimum Nanomaterial Characterization Requirements for Use in Toxicology Studies (MINChar Initiative, 2008)

What does the material look like?

Particle size/size distribution

Agglomeration state/aggregation

Shape

What is the material made of?

Overall composition (including chemical composition and crystal structure)

Surface composition

Purity (including levels of impurities)

What factors affect how a material interacts with its surroundings?

Surface area

Surface chemistry, including reactivity, hydrophobicity

Surface charge

Overarching considerations to take into account when characterizing engineered nanomaterials in toxicity studies:

Stability—how do material properties change with time (dynamic stability), storage, handling, preparation, delivery etc? Include solubility, and the rate of material release through dissolution.

Context/media—how do material properties change in different media; i.e. from the bulk material to dispersions to material in various biological matrices? (“as administered” characterization is considered to be particularly important)

Where possible, materials should be characterized sufficiently to interpret the response to the amount of material against a range of potentially relevant dose metrics, including mass, surface area and number concentration.

As a result of these and other efforts, there is movement toward expecting and including detailed physicochemical characterization data in nanomaterial toxicology studies. This is becoming increasingly important as methodologies are developed to develop predictive models for engineered nanomaterials and—by extension—sophisticated materials in general. In 2006, Maynard et al. challenged the scientific community to work towards predictive models for nanomaterial impact (Maynard et al., 2006). Four years later, a number of initiatives are beginning to work toward this goal (Alvarez et al., 2009; Meng et al., 2009). These approaches depend on associating key material properties with mechanisms of biological interaction and ultimately effects. To be successful, they will depend on detailed physicochemical characterization of the materials under test.

However, knowing what needs to be measured is only part of the challenge, it is complemented by the need for tools to make appropriate measurements. Here, progress is still lacking, with even the minimum characterization requirements proposed by initiatives like MINChar challenging toxicologists. Three challenges in particular face the toxicology community as increasingly sophisticated materials are developed: presenting samples to test systems that are well-characterized, evaluating key physicochemical properties in situ, and developing analytical techniques that can provide useful information into tools that are accessible to the toxicology community. Each comes with its own challenges and it is likely to be many years before substantial progress is made. Yet evaluating and quantifying the toxicology of sophisticated materials will depend on new tools and methodologies in each of these areas.

One final significant challenge exists here: developing an understanding of the tolerance within which physicochemical characteristics show similar or markedly different biological behavior (National Academies, 2009). Most sophisticated materials will be manufactured within a certain range of physicochemical properties, ensuring functionality is achieved without resulting in over-costly production processes. As a result, materials will demonstrate variation in particle size, shape, surface properties, and other characteristics, both between batches and within samples. Understanding the association between variations in physicochemical characteristics and toxicity will be essential in developing the tools and methodologies to quantify and address risks presented by sophisticated materials. Central to this are three questions: (1) How precisely do physicochemical characteristics need to be measured? (2) What constitutes a significant change in characteristics such as particle size, shape, or composition when evaluating toxicity? and (3) How should the hazard associated with a material representing a distribution of physicochemical characteristic be evaluated? So far, very little progress has been made toward addressing these issues.

Biointeractions

Over the past few years, concerns have been expressed in the lay press and elsewhere about the “new toxicities” of nanomaterials as they interact with biological systems. These concerns appear to be based on two potential types of interaction: (1) emergent quantum mechanical properties of nanomaterials may lead to novel interactions with biology and (2) the matching of scales between biological machinery and the engineered nanomaterial may lead to new mechanisms of interaction. On the first assumption, little information exists in the peer-reviewed literature that suggests a direct quantum mechanical interaction or energy transfer between engineered nanomaterials and biological systems in the absence of external energy sources. Specifically, in the absence of pumping of laser radiation deep into tissues (and thus activating size-mediated radiation-particle-biology interactions), it does not appear that such biophysical interactions have been detected in carbonaceous and other materials currently in high volume production.

Nevertheless, smaller quantities of new materials have been recently developed that allow for interesting biophysical interactions at this radiation/material/biology interface. In addition to the well-known photodynamic production of ROS for the treatment of cancer and the eradication of microbial biofilms, etc., gold nanoshells have been used for thermal ablation of tumor cells in deep tissues (Guerrero et al., 2010; Halas, 2010; Orringer et al., 2009; Wu et al., 2009). More recent developments in photonics have enabled the synthesis of quantum-dot like structures that enable multiphoton excitation and emission modes that may or may not produce ROS (Pecher et al., 2010). All these approaches require the injection of a relatively large tuned photon flux (usually generated by a laser) into the system. Whether or not the results are purely energetic interactions between engineered nanomaterials and biological macromolecules or whether there are further significant mechanisms of interaction remains to be determined. In this respect, biophysical measurements will have to be examined for toxicological plausibility in eliciting a pathological or pathophysiological outcome.

The second, more widely accepted mechanism is that nanomaterials are at the scale of biological macromolecules and lend themselves to hybridization. Indeed, macromolecules such as DNA have been used to solubilize highly hydrophobic single-walled carbon nanotubes (Yamamoto et al., 2010). It is this “matching of scales” that permits nucleic acids, lipids, proteins, and likely other macromolecules to interact with and coat the surface of engineered nanomaterials to produce harm. For example, single-walled carbon nanotubes have been shown to reproducibly gain access to cell nuclei and induce DNA strand breaks or formation of micronuclei (Cveticanin et al., 2010). Perhaps it is no surprise that nanomaterials are recognized by biological macromolecules and are adsorbed the available surfaces. This adsorptive process is frequently linked to the early stages of inflammation and is collectively known as opsonization. The question remains whether nor not engineered nanomaterials will present new “antigenic challenges” that will enhance or diminish immunologic function.

Inflammation is a common biological initiator or promoter of numerous pathophysiological states that end in either acute inflammatory disease (e.g., formation of granulomata) or act as a subchronic/chronic initiator/promoter of more pernicious degenerative or neoplastic disease. The long-known immunological process of opsonization of microbial and other external intruders has recently been shown to play a critical role in mediating a variety of idiopathic, environmental, and iatrogenic disorders. The complement system is composed of a tightly regulated complex of proteins that enable host immune responses to foreign materials. These proteins adhere to the surface of foreign bodies, are themselves activated, and stimulate removal of the offending article(s) by phagocytes. Inappropriate activation of this system has been linked to a variety of pathophysiological states including asthma and lupus (Gonzalez et al., 2010; Sarma and Ward, 2010; Silva, 2010).

The seminal work of Dawson and Colleagues in Ireland has led to the firmer understanding that characterization of the chemical and physical nature of nanomaterials is simply a necessary first step to determining the potential of the material to subvert biological processes. The physicochemical nature of the nanomaterial is augmented by the adhesion of a layer of proteins, lipids, and lipoproteins that forms upon first contact with the biological milieu (Lynch et al., 2007, 2009; Hellstrand et al., 2009; Walczyk et al., 2010; Vauthier et al., 2009). It is, in fact, this entire complex of nanomaterial encased in biological macromolecules that cells encounter and to which they respond. The picture is further complicated by the fact that there is dynamic exchange of many macromolecules between the external space and the surface of the nanomaterial that is dependent on the anatomic location of the material. The sheer complexity of the surface dynamic provides for at least three biological outcomes: (1) the interactions between biological fluids and matrices may defeat intended targeting regimens that are designed to deliver highly functional engineered nanomaterials to specific anatomic sites and may increase off-target effects, (2) the adsorption of elements of the complement cascade may activate or inhibit inflammatory processes that would in the normal run of things enable the body to effectively dispose of biological and other environmental threats at the nanoscale, or (3) the nanomaterial itself may provide a stable platform for the inappropriate delivery of bioactive molecules to anatomic sites and initiate cell signaling processes that trigger an adverse (or beneficial if appropriately designed) event (Dawson et al., 2009; Gaucher et al., 2009; Moghimi et al., 2010; Serda et al., 2009). Given that many of the engineered nanomaterials that are in current mass production are stable and are likely to persist in the body, the dynamic nature of the interactions between the nano- and biological interface add to the complexity of what needs to be known in assessing the likely biological outcome (Alexis et al., 2008).

The latter consideration, which is modulated by the durability of nanomaterial, evokes the consideration of time as a dose metric. In addition to the descriptions of surface reactivity, chemical composition, charge, aspect ratio, etc., the cautious toxicologist will need to have a firm understanding of the ability of the material to be disposed of by the biological system in which it dwells (Kagan et al., 2010a). The work of Kagan et al. remains to be confirmed in more complex systems that includes tangled carbon nanotubes or carbonaceous nanomaterials that have been functionalized. Nevertheless, this study provides an intriguing insight into the possibility that there may exist for the most stable nanomaterials a dynamic equilibrium between those biological processes that induce accumulation of the material and may confer harm and those that ultimately dispose of the material in situ (Kagan et al., 2010b). In its ultimate redux, the kinds of analysis that are required for adequate hazard and risk assessment of engineered nanomaterials specifically and increasingly sophisticated materials more generally are far more extensive and complex than those required for traditional chemicals. Success in preventing long-term harm and protect public health will require engagement in multidimensional high-content analyses of the pertinent ’omics with reference to traditional metrics such as dose, duration, and hazard. The general toxicity or off-target effects of emerging sophisticated materials will require investigation through selective profiling against large panels of potential targets. At present, these kinds of studies are expensive and their utility in estimating the risk of harm is not well characterized. Unless the experimental design is tailored to take into account known biological processes and is constrained to pathways that are known to induce identifiable harm (not merely a change in expression), the implementation of such approaches will not be widely adopted, and our ability to improve our collective predictive capabilities will be severely diminished (Merino et al., 2010).

A particular challenge associated with understanding interactions between nanomaterials and biological systems is that of relating in vitro observations to in vivo behavior. Recent research has begun to push the capacity of in vitro approaches to rapidly screen nanomaterials for potential toxicity and to develop potential causal relationships between characteristics and biologically relevant behavior (George et al., 2010; Meng et al., 2009). Yet toxicology studies in vitro carry with them several limitations, confounding factors, and caveats that make more complicated the incorporation of such data in the assessment of the risk potential of nanomaterials. For example, many published experiment apply a suspension of well-characterized nanomaterial to cell cultures with little consideration of the way in which particles behave in fluid suspensions. In addition to the well-known behavior of agglomeration and aggregation in aqueous environments, particles show a variety of behaviors that include surface adsorption, changes in surface charge and charge distribution, and nonuniform spatial distribution. Indeed, the latter may result in the delivery of widely differing amounts of nanoparticle to the surface of cells in culture (Teeguarden et al., 2007). This simple physical interaction of the nanoparticle with the suspending medium brings into sharp focus a variety of other considerations that cause an apparent difference in potency in vitro.

On the biological side, deficiencies in the ability of the cells in culture to recapitulate their phenotype in the context of the intact tissue are well known and range from the loss of biochemical functions critical to the biotransformation/bioactivation of xenobiotics to the morphology of the cell. Also, because most in vitro toxicologic investigations use monocultures of epithelia, in vitro culture models cannot engage in paracrine signaling between, for example, mesenchymal elements of the native tissue or dendritic cells that themselves take part in humoral signaling with other lymphoid tissues. The latter has recently been shown to be important in whole-body responses to nanomaterials (Mitchell et al., 2007). Interestingly, there have been exciting developments in the development of nanomaterials for enhanced cultured cell models that aid in the preservation of physiologic and morphologic function. These nanomaterials show considerable promise for neurotoxicology and neuroscience, for example, and include nanofibers such as polycaprolactone in the promotion of neuronal differentiation and orientation of neuronal progenitor cells derived from human embryonic stem cells (Mahairaki et al., 2010). Similarly, electrospun bioactive nanomaterials that mimic the properties of extracellular membranes have been developed and successfully deployed in the culturing of human renal tubular cells in a format that recapitulates the intercellular tight junctions and initiates formation of the brush border and expression of biologically relevant transport mechanisms such as γ-glutamyl transpeptidase (Dankers et al., 2011). Even newer materials have been developed that permit the dynamic inward budding of membranes (exosomes) that encapsulate small RNA's in a format ready to deliver to cultured and other cell types (Pegtel et al., 2010; Zomer et al., 2010). Undoubtedly, many of these advances and others in nanotechnology/nanomaterials will be useful in addressing and overcoming some of the complicating factors and limitations of cell culture in toxicologic research. There are on the drawing board wide variety of nanotechnologies for medical imaging and therapy. Overwhelmingly, the new nanomaterial-based approaches to therapy and imaging use nanoparticles that aim to improve the pharmacokinetic/pharmacodynamic profile of existing (predominantly hydrophobic) compounds. In that sense, these are “smart excipients” or “smart carriers.” An advanced embodiment of the smart carrier approach is the targeting of highly branched dendrimers of various chemical types and incorporate different cyclic cores such as carbopeptides, carboproteins, octopus glycosides, inositol-based dendrimers, cyclodextrins, calix[4]arenes, resorcarenes, cavitands, and porphyrins (reviewed by Sebestik et al., 2010). Even the “simple” liposome has benefitted from advances synthesis at < 100 nm and the ability to perform sophisticated surface chemistry (reviewed by Jølck et al., 2010). Hydrogels of the kind described elsewhere in this review have also been deployed for the imaging a treatment of tumors. Perhaps not surprisingly, the second generation of nanotechnologies for medicine is already under development and have been funded by the NIH Common Fund (http://nihroadmap.nih.gov/nanomedicine/fundedresearch.asp). These technologies are frequently inspired by the kind of solutions provided in the efficient nanomachines of nature, for example, cellular machinery that packages proteins with immensely massive forces, nanomachines that sense photons and wavelengths outside those visible to the human eye, etc. The use of newer (macro) technologies that work at the nanometer scale, for example, photonic tweezers, femtosecond pulsed lasers, and ultrasound, enable the development of tunable molecular assemblies that convert imparted energy into beneficial mechanical work aimed at repairing defects within the cell. All these advances will, as they are deployed in medicine and beyond, impose massive strain on the current toxicologic paradigms used in the evaluation of chemicals and the first generation of nanomaterials.

Considering further the intersection between beneficial use, biointeractions, and potential biological impact, the pharmaceutical industry has used a variety of smart excipients for targeted drug delivery for many years that dissolve with time and give rise to soluble chemicals that are readily excreted in the urine and/or feces thereby reducing the potential for bioaccumulation and off-target effects. Other approaches use the dissolution of the nanomaterial to reduce complement-mediated activation following admission into the vascular space (Gao et al., 2008). The picture is more complicated when the surface of the material is coated with targeting ligands for delivery of the nanodevice to a specific anatomic location or cell type. The surfaces of copolymer blends of polylactic-glycolic acid and polyethylene glycol nanoparticles have been successfully grafted with a number of ligands such as herceptin (anti-HER2 antibody). The nanoparticle itself may be loaded with an anticancer drug or a smaller nanoparticle such as nanosilver and permits delivery of therapeutics to a variety of cells, tissues, or surfaces. The targeting effect has been shown to be quantitatively controlled by two major approaches: (1) adjusting the copolymer blend ratio of the nanoparticle matrix with concomitant changes in the geometry of the surface and alterations in ligand density on the surface of the nanoparticle surface, and (2) adjusting the molar ratio of herceptin to available free amines appearing on the nanoparticle surface. Both these approaches (within limits) permitted a linear relation between the concentration of nanoparticles administered and the amount binding to tumor cells (Liu et al., 2010; Shameli et al., 2010). Of course, in the development of a nanomedicine, the purpose is to develop a material that is a priori biocompatible. With the notable exception of the nanomaterials for devices and prostheses, the synthesis of new therapeutic agents is rarely guided by considerations such as hardness, tensile strength, ability to absorb blunt/sharp force trauma, etc. These factors are more frequently the domain of consumer products and manufacturing.

The deployment of nanomaterials and other sophisticated materials in medical and consumer products is still in its infancy. As such, there exists a fleeting opportunity for a cross-over between these two disparate domains and sharing of pre-competitive data on hazard and exposure that may catalyze the smart design of materials that minimize harmful bio-nano-interactions and maintain the desired physicochemical properties imparted by a scale less than 100 nm.

LOOKING TO THE FUTURE

Nanotoxicology is a field that has been propelled into the limelight by science, speculation, and a growing push toward developing new and unusual materials in products. Its grounding in science is clear—research shows that the size, shape, chemistry and other physicochemical parameters of physical objects affect how they interact with biological systems and the potential impacts they have. Yet the uniqueness of the field in terms of the “nano” prefix is perhaps not so clear. Size matters—this is indisputable. But decades of research have demonstrated that particle size matters at the micrometer scale when it comes to human health, as well as at the nanoscale, raising the question of whether nanomaterial toxicity can be understood as an extension of what we know about larger scale materials or whether there is something unique about how nanoscale materials interact with biology that justifies them being singled out? Research over the past 20 years has shown that nanoscale materials can show unexpected and unusual toxicity and that physicochemical complexity is an important mediator of toxicity at the nanoscale. But it also shows that we are far from developing a precise understanding of how these parameters govern mechanisms of interaction or how they are empirically associated with response. In effect, the form and chemistry combined of materials entering the body are important, but the indicators of emergent risks are more complex than a simple size range.

The challenges of understanding how form and chemistry mediate toxicity will become increasingly apparent as emerging technologies lead to increasingly sophisticated materials. This is critical where ever-more complex and multifunctional therapeutics are being designed at the nanoscale for introduction to the human body. But it is also important where sophisticated materials may enter the body via other routes—as components of food, or through nonintentional exposure during product manufacture, and use, or during and after disposal. Advances in materials science, synthetic chemistry, biotechnology, and other areas are leading to materials that demonstrate designed and adaptable functionality; that obscure distinctions between biological and nonbiological substances; and that have multiple and often intricate components. Future sophisticated materials are more likely to resemble the complexity of human-scale engineered devices, rather than the simplicity of unique chemical entities. And this in turn means that a more sophisticated and systems-based approach to assessing their toxicity and addressing their potential risks is needed.

In moving forward, the risk assessment paradigm remains relevant. Ultimately, decisions need to be driven by a science-based approach that links dose to response. However, substantial work is needed in applying the risk assessment paradigm to new materials, so that a science-informed understanding of how to quantify and predict the potential risks of sophisticated materials is developed. At the same time, there is a parallel and somewhat intertwined challenge: Quantitative toxicology and risk assessment are unlikely to keep pace with the accelerating development of emerging sophisticated materials, meaning there will be a growing knowledge gap between the materials being produced, and the knowledge needed to ensure their safe use. Bridging this gap will require new approaches to evaluating risk and making decisions in the face of potential risks where there is incomplete information on exposure, hazard, and response. And given the dynamic nature of emerging materials as they are generated, used, disposed of and recycled, these approaches will need to be established within a life cycle framework (Som et al., 2010). Developing models that predict associations between sophisticated materials, biological interactions, and impacts will be a critical part of this (Maynard et al., 2006). There will also be an increasing need to push risk assessment—and thus toxicology—upstream in the innovation process, allowing early decisions to be made on safe and responsible product development (Owen et al., 2009). In effect, a new science of risk is needed that draws together the physical, biological, and social sciences to develop an integrated approach to the emerging challenges presented by sophisticated materials.

Since publication of Oberdörster's and Ferin's research on the size-mediated response to inhaled TiO₂ particles over 20 years ago, our understanding of how physicochemical characteristics mediate material toxicity has grown by leaps and bounds (Ferin et al., 1990; Oberdörster et al., 1990). We can now begin to appreciate the challenges presented by simple nanoscale materials such as TiO₂, ZnO, Ag, carbon nanotubes, and CeO₂. But these simple materials are merely the vanguard of a new era of complex materials, where novel and dynamic functionality is engineered into multifaceted substances. If we are to meet the challenge of ensuring the safe use of this new generation of substances, it is time to move beyond “nano” toxicology and toward a new toxicology of sophisticated materials.

FUNDING

The University of Michigan Risk Science Center (to A.D.M.); ES 2R01 ES08846 (to M.A.P.).