Chapter 28: Nanotoxicology

Since the classic talk by Richard Feynman (1959) entitled “There Is Plenty of Room at the Bottom,” nanotechnology has grown to a multibillion dollar industry worldwide, with 1300 nanotechnology-enabled products in commercial use by 2010 (Woodrow Wilson Center, 2012). The potential of adverse effects from exposure to “nanophase materials” was already pointed out earlier (Oberdörster and Ferin, 1992; Oberdörster et al., 1992), and concerns about human and environmental health and safety of engineered nanomaterials (ENMs) were initially raised in 2003 (Colvin, 2003). Since then, toxicity of high volume, commercial nanomaterials including nanosilver, fullerenes, quantum dots, carbon nanotubes (CNTs), and metal oxide nanoparticles (NPs) have been summarized in several reviews (Borm et al., 2006; Nel et al., 2006; Donaldson et al., 2004; Boczkowski and Hoet, 2009; Krug and Wick, 2011; Kunzmann et al., 2011). New ENMs and composites are continually emerging with potential for significant commercial applications in energy generation, environmental sensing and remediation, aerospace and defense, and medical diagnosis and therapy. Examples of nanoscale materials of different shapes and sizes are depicted in Fig. 28-1. Investigation of the magnitude of release of manufactured nanomaterials and their subsequent fate, transport, transformation, and potential for human and environmental exposure and toxicity (Fig. 28-2) is an urgent priority (Mueller and Nowack, 2008).

The National Nanotechnology Initiative (NNI, http://www.nano.gov/) defines nanotechnology as the understanding and control of matter at the nanoscale at dimensions between approximately 1 and 100 nm, where unique phenomena enable novel applications. Roco (2005) defined the sizes as ranging from the intermediate length scale between a single atom or molecule and about 100 molecular diameters or about 100 nm, and emphasizes the ability to measure and transform at this nanoscale and exploiting the properties and functions specific for the nanoscale.

The European Union (EU Commission, 2011) defined nanomaterial as a natural incidental or material containing particles as free, aggregates, or agglomerates, with 50% or more (by number) of them in the number size distribution in the range 1 to 100 nm of one or more of the external dimension, and expressed specific concerns for environment, health, safety, and also competitiveness. Because some nanoscale materials are less than 1 nm, it was clarified that fullerenes, graphene flakes, and single-walled carbon nanotubes (SWCNTs) below 1 nm are considered nanomaterials.

The continuing introduction of nanoscale materials into consumer products, such as TiO₂ in sunscreen creams, antibacterial Ag in textiles, multiwalled carbon nanotubes (MWCNTs) in sport equipment together with increasing numbers of publications reporting toxic responses mostly observed in in vitro studies, has led to increasing concern and more public awareness about potential adverse health effects, fostered particularly also by alarmist headlines in the popular press. Unrealistic high doses as well as questionable study designs are most often to blame. Thus, questions have to be raised about real versus perceived risk of nanotechnology applications, what is hype, what is reality? To provide answers to these questions is one of the goals of the subdiscipline of nanotoxicology.

Nanotoxicology can be defined as the study of adverse effects of nanomaterials on living organisms and the environment. As illustrated in Fig. 28-2, there are multiple occasions for exposures to nanomaterials of humans and the environment and there is a serious need for investigation of the potential to cause adverse effects. The need of more toxicological research with the goal of characterizing risks associated with nanotechnology becomes even more urgent when considering the many beneficial and promising applications of nanotechnology for the betterment of human life, in particular also for prophylactic, therapeutic, and diagnostic medical applications. The applications of nanotechnology for human health defines the new field of nanomedicine, and the development of new nano-enabled medical technologies must be accompanied by nanotoxicity studies to characterize undesirable effects and ensure the safety of new therapies and medical devices.

The field of nanotoxicology evolved from research of ambient airborne ultrafine particles (UFPs), particles up to about 100 nm from diverse anthropogenic and natural emission sources, including tailpipe emissions, power plants, volcanic eruptions, forest fires, and others, as summarized in Table 28-1, which gives examples of three classes: natural NPs, unintentional anthropogenic NPs, and intentional or engineered anthropogenic NPs. The same toxicological concerns and problems need to be addressed for UFPs and ENMs, they require similar research strategies with respect to study design and characterizing effects and underlying mechanisms at the primary portal of entry and in remote secondary organs. However, not only well-established respiratory tract and air pollution toxicology (see Chaps. 15 and 29) form the basis for nanotoxicology, but findings and knowledge accumulated in other traditional fields of research provide useful complementary information and foundations for nanotoxicology. For example, virology (nanosized structures–cell interactions), metal and polymer-fume toxicology (small particle size and systemic effects), silica toxicology (particle surface reactivity), fiber toxicology (particle shape and biopersistence), radionuclide toxicology (chemistry and biokinetics), and lung deposition studies/modeling (respiratory tract dosimetry) (Oberdörster et al., 2007a,b). Input from these research areas will help in the design of studies and interpretation of results of nanotoxicological research. For example, a major challenge is to correlate the distinct physicochemical characteristics of engineered nanostructures with their toxicological behavior, as will be discussed later.

Of utmost importance for nanotoxicology, as for toxicology in general, are dose-related issues. In particular, administered doses are mostly extraordinarily high when compared to doses achieved in vivo under realistic exposure scenarios. Slikker et al. (2004) stressed that the dose-dependent mechanisms are due to saturation processes underlying toxic responses between typical high acute experimental doses and lower human doses from realistic exposures (Fig. 28-3). Although high doses are justifiable for hazard characterization, ignoring real-world exposure situations may lead to erroneous conclusions about associated risks and causative mechanisms. Thus, consideration of the impact of dose, dose rate (influenced by dosing method in vitro and in vivo), and dosimetry should always be given highest priority when assessing nanomaterial toxicity. Dosimetry involves quantifying by measuring or calculating the amount of nanomaterials per body, per organ, per cell (microdosimetry), tissue or airway generation (inhalation route of exposure). Dosimetry should not be confused with “dosemetric,” which is defining a dose in terms of an inherent property (physical, chemical, reactivity) of nanostructures. Neither should dosimetry or dosemetric be mixed up with “exposure.” These issues will be discussed in more detail later.

The goals of nanotoxicology are to identify and characterize a hazard of ENMs for purposes of risk assessment for humans and the environment require a highly multidisciplinary team approach, covering expertise in toxicology, biology, chemistry, physics, material science, geology, exposure assessment, physiological-based pharmacokinetic (PBPK), and fate and transport modeling, and medicine is necessary to develop testing strategies, establish toxicity ranking, determine “safe” exposure levels, and derive preventive exposure guidelines. As depicted in Fig. 28-2, exposures throughout the lifecycle of ENMs, from their source to their disposal have to be considered. Dispersion and intermediate transformations in air, water, food, and soil are important modifiers of ENM—receptor interactions. Identifying underlying mechanisms of such interactions will aid in assessing risk and establishing preventive measures. Developing predictive models is a distant goal for both human and eco-nanotoxicology.

Perspectives: Engineered Nanoparticles Versus Ambient Particulate Matter

A large body of publications on the toxicology of airborne ambient particulate matter (PM) has shown that they can elicit adverse effects not only in the respiratory tract but also in secondary organs and systemically (see Chap. 29 on Air Pollution). An extensive critical review of the health effects of fine particulate air pollution (Pope and Dockery, 2006) pointed out that exposure to the smallest fraction of PM, referred to as UFPs or PM_0.1, particles smaller than 0.1 μm or 100 nm, has been associated with effects in the cardiovascular and central nervous system as a consequence of their translocation to and distribution via blood circulation and neurons. It is noteworthy that the definition of UFPs is based on the same maximum dimension (100 nm) as the US federal definition of nanotechnology. As will be discussed later, translocation of UFP and engineered nanoparticles (ENPs) has been clearly demonstrated, but only small amounts are translocating, and thus direct interaction of the translocated particles with organs beyond the lung may not be the sole explanation for their effects. Epidemiological studies have demonstrated that a major factor for inducing adverse effects from ambient particulate air pollution—including UFP—is susceptibility such as pre-existing disease (asthma, cardiovascular disease, diabetes), age (very young, elderly), or genetic background (polymorphism).

Properties and Behaviors of ENPs Versus Larger Particles

Table 28-2 contrasts differences between NPs (<100 nm) and larger particles (>500 nm) in terms of some general characteristics, translocation propensity, interactions with cells and effects, assuming inhalation exposure, and the respiratory tract as the portal of entry. Biological systems do not perceive a precise boundary at the 100 nm size threshold, but rather a gradual transition between nano- and larger-sized particles, and therefore, the 100 to 500 nm size is shown but not covered in the table.

Marked differences—which will be discussed in the following sections—between the two size classes are as follows: The high number and surface area per unit volume, which makes them chemically more reactive (eg, as catalysts) and similarly increases their biological/toxicological activity toward cells and tissues; the underlying mechanisms for deposition of inhaled particles in the respiratory tract, which is governed by diffusion or Brownian motion for NP, and by sedimentation, impaction, and interception for larger particles; and there is a difference with regard to the extent and impact of adsorption of proteins and lipids (corona formation) at the primary portal of entry and secondary target sites.

Efficient mucociliary and alveolar macrophage-mediated elimination following deposition in the respiratory tract is efficient for both nano- and larger particles once they are internalized by macrophages. NPs inhaled and deposited as singlets, however, are too small to be efficiently recognized and phagocytized by alveolar macrophages (Oberdörster et al., 2005; Semmler-Behneke et al., 2007; Geiser et al., 2008)—unless they are aggregated or agglomerated to form larger particles—and thus overall alveolar macrophage-mediated clearance in the lung is poor. In contrast, uptake by epithelial cells and translocation into blood and lymphatic circulation occurs regularly for NPs, and only under heavy overload conditions for larger particles. Uptake of nanosized particles into sensory nerve endings of the tracheobronchial tree has also been described (Hunter and Undem, 1999), which is unlikely to occur for large particles.

Differences in cell entry mechanisms between NPs and larger particles and intracellular distribution to mitochondria and the nucleus are discussed in subsequent sections. With regard to the type of effects induced at the primary site of entry there does not seem to be a fundamental difference. However, effects in secondary organs are more likely to develop, possibly aided by a direct effect of translocated NP.

Manufactured NPs show a range of other size-dependent properties and behaviors. Among these are depressed melting points, size-dependent electronic band gaps, a quantum mechanical effect that gives rise to size-tunable fluorescence in quantum dots (semiconductor NPs), and confined plasmon resonance, which shifts the light absorption and scattering in a way that alters the colors of nanoscale materials from their larger-sized counterparts giving—for example—rise to nanosilver particles that are yellow and gold NPs that appear red. Unique size-dependent properties of NPs form the basis for many of their technological uses, but their significance for biological effect still needs to be evaluated.

Classes of ENMs

Manufactured nanomaterials have an enormous range of chemical composition, geometry, and complexity ranging from simple isometric forms (NPs), one-dimensional (1D) forms (fibers or tubes), and two-dimensional (2D) forms (plate-like or disk-like materials). A large fraction of the stable elements in the periodic table have now been cast into NPs, or incorporated as part of a nanocomposite structure. Nanomaterials can be manufactured from metals, metal oxides, sulfides or selenides, carbon, polymers, or from biological molecules including lipids, carbohydrates, peptides, proteins, and nucleic acid oligomers. Fig. 28-4 presents a nanomaterial classification system based on a matrix involving geometry (particles, 1D forms, 2D forms) and chemistry (metals, semiconductors, ceramics, carbons, polymers; Tran and Nguyen, 2011; Rosenthal et al., 2011; Cassee et al., 2011; Feng and Liu, 2011; Duncan, 2011). The combination of geometry and materials chemistry gives rise to vast number of potential nanomaterials. Not all will be commercially significant, but many are being synthesized and characterized in R&D laboratories around the world. Nanomaterials may be applied to surfaces such as biomedical implants to enhance their function and biocompatibility, incorporated into nanostructured solids or composites to improve strength, conductivity, and durability, and fabricated into complex, active structures for chemical or biological sensors or other devices.

There are many approaches for synthesizing and manufacturing nanomaterials. Some methods use high-temperature processing in the vapor phase, such as inert gas condensation, flame synthesis, or “floating catalyst” chemical vapor deposition (Zhang et al., 2011). These processes produce dry powders that can lead to occupational exposures when reactors are open or the powdered materials are handled or transferred. Other methods such as physical or chemical vapor deposition grow nanostructures on substrates. Substrate-bound nanostructures are less likely to cause exposures of concern, but can become detached during processing or use. A large class of methods uses solution processing, also known as “colloidal synthesis” (Tran and Nguyen, 2011). These methods involve reagent addition to liquid solvents to nucleate NPs that remain suspended in the liquid. This class of methods is less likely to lead to inhalation exposures, but they can occur if the liquid is subjected to ultrasonication, which can produce mists, or if the NPs are dried into powders or transferred onto substrates and dried for further processing (Hallock et al., 2009).

Finally, most nanotoxicology research has focused on a limited set of nanomaterials with high manufacturing volume or the potential for high manufacturing volume in the near term (including ZnO, TiO₂, fullerenes, CNTs, nanosilver, quantum dots). Increasingly, the nanotoxicology field will have to focus on the other materials under development, especially as technological applications for those materials approach commercialization. It will be a major challenge for the nanotoxicology field to acquire the data necessary for risk assessment on the many new nanoproducts that are anticipated over the coming decades. Finally, as nanoscience progresses, there will be less emphasis on simple geometries and chemistries (see Fig. 28-4) and more emphasis on complex material structures that combine nanoelements into active or smart structures (Dvir et al., 2011). Early examples of this are the synthesis of two-component or “dumbbell-shaped” NPs composed of gold and iron oxide (Xu et al., 2007) or iron-platinum nanowires (Wang et al., 2007). Understanding the toxicological profile of complex nanoassemblies will also present a major challenge to the field in the future.

Physicochemical Properties of Nanomaterials Relevant for Toxicity

Table 28-3 summarizes the nanomaterial properties known or thought to be relevant to biological responses. These properties are discussed in more detail in the section below, or in some cases in later sections on mammalian toxicity and ecotoxicity.

Surface Area and Reactivity

Nanomaterials display unique chemical and physical properties that can be exploited for novel consumer, industrial, defense, and medical applications, but these properties may also contribute to adverse environmental and health effects. At the nanoscale, chemical and physical forces govern the assembly of atoms or molecules resulting in nanostructures with a high surface-to-volume or surface-to-mass ratio. For example, 5 nm NPs have 1000 times the surface area of 5 μm particles of the same chemical composition and at the same airborne mass concentration. This high surface area of suspended NPs is responsible for increased surface reactivity, increased adsorption of chemicals, and strong catalytic activity. As particle diameter decreases below ~100 nm, the number of exposed surface molecules increases sharply (Fig. 28-5; Oberdörster et al., 2005). High surface reactivity is a desirable property for catalysis; however, the large number of exposed surface molecules or atoms exposes surface defects, vacant sites, and dangling chemical bonds that enhance chemical and redox reactivity. Thus, the NP surface has high surface area, high surface energy, and high surface reactivity relative to micron-sized particles of the same chemical composition (Fenoglio et al., 2011).

Surface Charge

NPs have a strong tendency to form aggregates or agglomerates in the dry state, and in many cases also in liquid suspension (Fig. 28-6). This tendency is a natural consequence of small size, which leads to intermolecular attractive forces that are strong relative to gravity or to other forces that disrupt aggregates. Attractive forces between particles are determined by size, density, chemistry, and surface charge. In water, hydrophobic forces drive the agglomeration of many NP types. To obtain stable dispersions of unagglomerated nanomaterials, it is typically necessary to add macromolecular coatings that prevent particle–particle attachment through steric hindrance, or to impart an electrical charge on the NP surfaces that lead to particle–particle repulsion. A common indicator of surface charge is the “zeta potential,” which is the electrical potential at the outer surface of a sphere that includes the particle and adjacent water molecules that travel with the particle during its motion. A common rule of thumb is that the zeta potential must be >30 mV or <−30 mV for repulsion to be sufficiently strong to avoid agglomeration. In physiological fluids or seawater, sodium chloride, and other ions “screen” the particle surface charge and lower the zeta potential, often causing aggregation. This property is a complicating factor in nanotoxicology assays (discussed below) and is a challenge for some biomedical applications. NPs may also adsorb molecules from their environment that increase stability (Nel et al., 2009). Following injection or inhalation, NPs become coated with biomolecules such as serum albumin, an amphiphilic protein that carries poorly soluble molecules in the blood, or phospholipids in lung surfactant lining the alveoli as discussed in detail subsequently. For drug delivery applications (De Jong and Borm, 2008), NPs may be coated with biocompatible surfactants such as sodium cholate, or by biocompatible polymers such as phospholipid-polyethylene glycol (PEG) that stabilize colloidal suspensions of NPs through steric hindrance—a mechanism separate from electrostatic repulsion as indicated by zeta potential.

Surface Chemistry

The unique physicochemical properties of nanomaterials responsible for their enhanced surface reactivity in biological systems are summarized in Fig. 28-7. Small size, high surface reactivity, and surface chemistry of NPs govern different types of interactions with biological molecules. High surface area and exposed surface atoms or molecules promote increased dissolution and release of ions from metallic or metal oxide NPs relative to bulk particles of the same chemical composition. For example, Ag⁺ ions can be slowly released from nanosilver particles in aqueous environments by oxidation (Liu and Hurt, 2010). This property accounts for much of the bactericidal properties of clothes, toys, and medical devices coated or impregnated with nanosilver particles. Metal ions are toxic to bacteria and aquatic organisms by inhibition of enzymes and transport proteins. Silver ions have been shown to induce swelling of the lamellae of trout gills due to inhibition of carbonic anhydrase and Na⁺/K⁺ ATPase activity (Scown et al., 2010). ZnO NPs are incorporated into sunscreens where they absorb ultraviolet (UV) light; however, in water, Zn⁺² ions are rapidly released and cause acute toxicity (Xia et al., 2008a). Surface hydrophilicity of charged NPs increases their ability to be suspended in water, whereas surface hydrophobicity of fullerenes or graphene repels water and enables these hydrophobic nanomaterials to partition into lipid membranes and gain entry into target cell (Sayes et al., 2005; Titov et al., 2010). TiO₂ NPs are used in sunscreens because they absorb UV radiation, but are too small to scatter visible light and appear transparent to the human eye. UV absorption can also lead to formation of electron/hole pairs that react on the particle surface to generate reactive oxygen species (ROS) such as superoxide anion (O₂^•-) from molecular oxygen (O₂) (Schilling et al., 2010). Surface defects, especially on crystalline NPs, expose electron donor/acceptor active groups that donate an electron to molecular oxygen also generating superoxide anion. Spontaneous or photoactivated generation of electrons accounts for the semiconductor properties of fullerenes and some metal oxide NPs (Xia et al., 2006).

CNTs are synthesized in the presence of metal catalysts, usually iron, nickel, or cobalt, and may contain redox-active metal residues that can undergo redox cycling and iron-catalyzed generation of highly reactive hydroxyl radical (OH^•) via Fenton chemistry (Kagan et al., 2006; Guo et al., 2007). Nanomaterials with high surface area can adsorb organic molecules such as polycyclic aromatic hydrocarbons that are potentially carcinogenic (Yang et al., 2006) and quinones that also participate in redox cycling and generation of free radicals (Nel et al., 2006). Cationic NPs that have surface amide groups and cationic dendrimers are especially cytotoxic because they induce membrane damage, especially in lysosomes where the “proton sponge effect” leads to accumulation of water and chloride ions and osmotic rupture (Xia et al., 2008b).

In summary, surface properties are major determinants of biological reactivity due to (1) high surface area, (2) surface charge, (3) hydrophobicity and partitioning into lipid membranes, (4) dissolution and release of metal ions, and (5) redox activity leading to generation of ROS (Nel et al., 2006; 2009; Xia et al., 2008b; Stone et al., 2009).

Unique Quantum and Magnetic Properties

NPs also display unique quantum and magnetic properties relative to micron-sized particles. For example, noble metal NPs show preferred wavelengths for absorption and scattering of light due to resonant electron oscillations. This “confined plasmon resonance” phenomenon is shown by 40 to 100 nm diameter gold or silver NPs and leads to color shifts that impart a yellow color to nanosilver and a red color to some types of nanogold (Tran and Nguyen, 2011). Ferromagnetic NPs (eg, iron, nickel, cobalt, and some oxides) less than about 10 nm in diameter exhibit the property of superparamagnetism. These particles respond strongly to an external magnetic field but have no permanent magnetic moment in the absence of the field because thermal energy is sufficient to cause fluctuations in the small magnetic dipole associated with each particle (Cole et al., 2011). This property is exploited for contrast enhancement in diagnostic MRI and for hyperthermia induced by an external magnetic field to kill tumors targeted by magnetic NPs (Fortin et al., 2007).

Geometry and Dimensions

The geometry and dimensions of nanomaterials are also important in cellular uptake, systemic translocation, and potential toxicity as described in detail (Section III). Nanomaterials can enter target cells by passive diffusion, direct physical penetration, or active, receptor-mediated uptake by endocytosis or phagocytosis depending on their size and extent of agglomeration. For example, small NPs up to 50 nm in diameter appear to enter cells and subcellular organelles, including mitochondria and the nucleus, by passive diffusion (Geiser et al., 2005; Chithrani, 2010). SWCNTs have been shown to directly puncture bacterial cells leading to osmotic lysis and death (Kang et al., 2007; Akhavan and Ghaderi, 2010). SWCNTs have been reported to translocate from the alveoli into the interstitium of the lung where they promote collagen deposition and interstitial fibrosis (Wang et al., 2011a). Rigid, long, high aspect ratio nanomaterials, similar to amphibole asbestos fibers (Palomaki et al., 2011), are recognized by macrophages; however, they undergo incomplete uptake or frustrated phagocytosis (Shi et al., 2011) resulting in impaired clearance from the lungs, lysosomal disruption, and activation of the inflammasome resulting in release of proinflammatory cytokines (Hamilton et al., 2009; Donaldson et al., 2010).

Biopersistence

Finally, biopersistence of ENMs is an important factor in their environmental and biological toxicity. Biopersistence is related to dissolution, which not only produces biologically active ionic species (as discussed above), but can also degrade the particle and eventually clear it from biological tissue or the natural environment. ZnO NPs are an excellent example of an ENM that can exhibit acute cellular toxicity due to rapid release of Zn²⁺ ions. The rate of dissolution of metal oxides is increased by natural organic matter in the aequous environment (Xia et al., 2008b); therefore, these NPs have low biodurability and it is predicted that they would not bioaccumulate in the environment. Following uptake by macrophages at the site of entry, metallic and metal oxide NPs reach the acidic endolysosomal compartment where release of metal ions is accelerated. Intracellular release of Zn²⁺ or Ni²⁺ ions, for example, is linked with significant acute toxicity to target cells in the lungs (Xia et al., 2008b; Liu et al., 2007; Pietruska et al., 2011). ZnO NP dissolution is very rapid and is considered unlikely to be biopersistent in the lungs and induce chronic lung disease. In contrast, poorly soluble metallic nickel NPs appear to induce sustained intracellular release of Ni²⁺ ions resulting in persistent inflammation (Lu et al., 2009) and activation of the HIF1-α signaling pathway in human lung epithelial cells (Pietruska et al., 2011) that has been linked to nickel carcinogenicity (Costa et al., 2005). Jakubek et al. (2009) reported that release of yttrium ions in trace levels from CNTs displaces Ca²⁺ from voltage-gated calcium channels with implications for neuronal activity. So far, no chronic rodent carcinogenicity assays have been conducted to assess whether inhalation of metal or metal oxide NPs can induce lung cancer or neurological disease.

In contrast, carbon nanomaterials are potentially biodurable raising concern that they could bioaccumulate in the environment or be biopersistent in the lungs and other organs following inhalation during occupational exposure or direct injection or implantation in biomedical applications (Bianco et al., 2011). Some commercial types of long, rigid, high aspect ratio CNTs have been shown to induce toxicological endpoints similar to carcinogenic asbestos fibers (Sanchez et al., 2009) as discussed in detail below. Biopersistence in the lungs and pleural or peritoneal spaces is an important physicochemical characteristic of asbestos and man-made mineral fibers associated with carcinogenicity (Bernstein et al., 2005). Several recent studies have shown that carboxylated SWCNTs do undergo oxidative degradation in the presence of the plant enzyme, horseradish peroxidase and H₂O₂ (Allen et al., 2008; 2009) or following exposure to phagolysosomal simulant fluid containing H₂O₂ to mimic the lysosomal compartment of macrophages (Liu et al., 2010c). Oxidized or nitrogen-doped MWCNTs also showed both exfoliation of the concentric tubes and shortening in the presence of horseradish peroxidase and H₂O₂ (Zhao et al., 2011). Oxidatively degraded SWCNTs did not induce lung inflammation or toxicity following pharyngeal aspiration in mice (Kagan et al., 2010) providing proof-of-principle for deliberate design of engineered CNTs that are biodegradable and less likely to induce acute or chronic disease following inhalation or injection for tumor imaging or drug delivery (Bianco et al., 2011). Graphene oxide nanosheets have also been shown to be degraded in the presence of horseradish peroxidase and H₂O₂ (Kotchey et al., 2011) or by Shewanella bacteria that chemically reduce graphene oxide to graphene (Salas et al., 2010). These studies provide novel approaches for biodegradation of ENMs and prevention of long-term adverse environmental and health impacts.

The high surface area of NPs provides a platform for adsorption of a variety of biological molecules including proteins, lipids, and nucleic acids. Dawson and his coworkers described in detail rapid binding of proteins in biological fluids to the surface of NPs that they defined as the “protein corona” (Lynch et al., 2007). The protein corona is a dynamic interface between the NP and its local environment that governs its initial interaction with target cells (Dutta et al., 2007; Walczyk et al., 2010). The interaction of nanomaterials with blood plasma proteins has been investigated in most detail due to its importance in drug delivery, circulation time, organ distribution, and clearance (Karmali and Simberg, 2011). Protein adsorption is related to surface charge, hydrophobicity, and radius of curvature; however, NPs of diverse chemical composition and structure (eg, iron oxide NPs, polymeric NPs, and CNTs) avidly adsorb albumin, complement, fibrinogen, immunoglobulins, and apolipoproteins (Nel et al., 2009). Protein adsorption and dissociation is a complex, dynamic phenomenon (Cedervall et al., 2007a,b). Quantitative techniques to predict and characterize protein adsorption have been developed (Xia et al., 2010). The consequences of protein adsorption to NPs are less clear; although, depending on the NP surface, proteins may denature resulting in loss of normal structure and function, or altered enzyme activity, or unfolding that exposes new antigenic determinants. NPs of diverse chemical composition also induce protein fibrillation or aggregation to form amyloid fibrils in an acellular system (Linse et al., 2007). An important potential pathological consequence of serum protein adsorption to NPs is binding of fibrinogen leading to formation of blood clots (Khandoga et al., 2010).

NPs that are inhaled into the respiratory tract or ingested into the gastrointestinal tract encounter a mucus layer that provides a natural barrier to penetration of particulates and microorganisms. Mucus is a complex glycoprotein composed of hydrophilic, dispersed regions and hydrophobic, globular domains. NPs may adhere to mucins causing enlargement of the pore size with increased susceptibility to penetration of microorganisms (Wang et al., 2011b; Shvedova et al., 2007). Smaller, charged NPs may be repelled by the hydrophilic domains and will not be able to penetrate the mucus layer (Lai et al., 2009; Cone et al., 2009; Wang et al., 2011b). Aquatic organisms (Klaine et al., 2008) and bacterial biofilms (Rodrigues and Elimelech, 2010) are surrounded by similar extracellular barriers that may prevent NP penetration. If inhaled NPs penetrate into the alveoli of the lungs, they encounter the lung lining fluid covering a thin film of surfactant lipids. Natural or synthetic lung lining fluid, which is a mixture of serum albumin and dipalmitoylphosphatidylcholine binds readily to a range of NPs and enhances their dispersion for use in toxicological assays (Sager et al., 2007; Porter et al., 2008).

ENPs also bind nucleic acids and have been proposed as gene delivery devices (Chou et al., 2011). Depending on their chirality, SWCNTs may show sequence-specific DNA binding (Tu et al., 2009). Graphene oxide nanosheets preferentially bind single-stranded RNA or DNA and protect it from enzymatic degradation (Lu et al., 2010). Small graphene oxide nanosheets can also intercalate into double-stranded DNA and induce DNA breaks in the presence of Cu²⁺ ions (Ren et al., 2010). Size-dependent binding of gold NPs into the major groove of DNA has also been reported (Schmid, 2008).

The emerging nanotechnology industry has generated fear and distrust among the public about unpredictable toxicity and unintended consequences of human exposure to ENMs (Bainbridge, 2002; Cobb and Macoubrie, 2004). Despite these concerns, there has not yet been an outbreak of human disease associated with exposure to ENMs, although potentially long latency periods between exposure and manifestation of effects require continued vigilance to prevent future problems. The mechanistic pathways associated with toxicity that have been discovered using various in vitro and in vivo assays are predictable based on the physicochemical properties of ENMs (Nel et al., 2009; Meng et al., 2009). Some of these toxicity pathways were identified by studying UFPs associated with ambient air pollution (Xia et al., 2006), whereas others are the consequence of intracellular release of toxic metal ions (Xia et al., 2008b). Oxidative stress due to direct generation of ROS at the surface of NPs or indirectly by target cells following internalization of NPs (Unfried et al., 2007) is a common mechanism responsible for toxicity of ENMs. The most vulnerable subcellular organelles and physiological functions that can be perturbed by exposure to ENPs are summarized in Table 28-4. Depending on the dose, duration of exposure, type of NP, and the target cell, the cellular responses may be minimal and reversible, involve activation of adaptive responses, or severe, leading to significant changes in cell structure and function that may culminate in cell death (Meng et al., 2009). Specific examples of toxicity endpoints induced by exposure to ENMs will be used to summarize our current understanding of mechanisms relevant for nanotoxicology. Most of the mechanistic studies use a variety of in vitro cellular toxicity assays and relatively high concentrations for 24 to 48 hours of exposure. As discussed in detail below, extrapolation from these short-term, high-concentration exposures in vitro to chronic, low-dose exposures in vivo is problematic. Given the wide range of ENMs currently in commercial production or in the early development stages, understanding toxicity pathways is essential for developing a rationale strategy for in vitro screening and prioritization for chronic animal testing based on new principles for toxicity testing in the 21st century (Simmons et al., 2009; Walker and Bucher, 2009; Silbergeld et al., 2011).

The cell wall of bacteria and the plasma membrane of eukaryotic cells are the initial barriers to penetration of NPs into target cells. Both gram-negative and gram-positive bacteria have thick cell walls composed of complex lipids and peptidoglycans, respectively. Bacteria do not actively take up particulates; however, several investigators have shown direct physical penetration of bacterial cell walls especially by SWCNTs (Kang et al., 2007, 2008; Liu et al., 2009), graphene and graphene oxide nanowalls (Akhavan and Ghaderi, 2010), and zinc oxide NPs (Liu et al., 2009). Carbon nanomaterials are proposed to act as “nanodarts” creating holes in the plasma membrane resulting in extracellular release of cytoplasmic contents as assessed by efflux of ribosomal RNA and decreased survival (Liu et al., 2009).

A wide range of NPs have been designed to facilitate delivery of imaging agents, genes, proteins, and drugs into mammalian cells: metallic and iron oxide NPs, silica NPs, quantum dots, biocompatible polymers, liposomes, micelles, and dendrimers (Chou et al., 2011). Target cells may recognize untargeted NPs that bind to surface scavenger receptors (Kanno et al., 2007; Hirano et al., 2008), lectin receptors (Zhang and Monteiro-Riviere, 2010), and Toll-like receptors that recognize hydrophobic domains (Seong and Matzinger, 2004). NPs can also be designed to target specific cell surface receptors triggering internalization by phagocytosis, macropinocytosis, or endocytosis mediated by clathrin, caveolae, or lipid rafts (Nel et al., 2009; Lajoie and Nabi, 2010). In order to facilitate gene, protein or drug delivery, NPs can be engineered to escape from endosomes or lysosomes by coating with pH-sensitive polymers, viral capsids, cations, or biodegradable carriers (Chou et al., 2011).

Eukaryotic cells can actively internalize a variety of ENPs using energy-dependent, receptor-mediated endocytosis or phagocytosis (Zhao et al., 2011). Following endocytosis or phagocytosis, NPs follow a well-defined intracellular pathway usually leading to fusion with lysosomes (Oh and Swanson, 1996). The kinetics and mechanisms of cell uptake vary depending on the target cell, size, shape, and surface modifications (Chithrani, 2010). Most NPs and microparticles are stored in membrane-bound vesicles, endosomes, or phagosomes; however, a subset of NPs including mesoporous silica NPs and superparamagnetic iron oxide NPs coated with chitosan undergo exocytosis or release from endothelial cells (Slowing et al., 2011) and macrophages (Serda et al., 2010). Exosomes are derived from sorting endosomes and multivesicular bodies; smaller NPs show greater exocytosis than particles in the range of 30 to 50 nm (Chithrani, 2010; Serda et al., 2010). Selective coating of NPs to favor delivery to exosomes can be exploited for drug delivery to vascularized tumors or to promote cell–cell transfer of genes and proteins (Slowing et al., 2011).

NPs that are recognized by surface receptors may activate cell-type-specific signaling pathways leading to cell proliferation or death by apotosis, stress-related signaling, or calcium-mediated signal transduction events (Unfried et al., 2007). Dysregulated intracellular calcium ion homeostasis may be the consequence of influx across a damaged plasma membrane permeability barrier or release of calcium ions from the major intracellular storage sites, mitochondria, and endoplasmic reticulum. Sustained elevation in intracellular calcium can cause cell death by necrosis (Xia et al., 2006).

In addition, small polystyrene NPs and metallic NPs and dendrimers have been shown to enter a variety of target cells by passive diffusion in an energy-independent process (Lesniak et al., 2005; Geiser et al., 2005). Following uptake by diffusion, these small NPs are not compartmentalized into membrane-bound vesicles or lysosomes but are found free in the cytoplasm where they may secondarily enter the mitochondria or even the nucleus.

Mitochondria are the site of aerobic respiration and generation of adenosine triphosphate (ATP). Small NPs have been observed inside mitochondria (Li et al., 2003; Xia et al., 2006) where they may disrupt electron transport in the inner mitochondrial membrane and generate excess endogenous ROS. Direct or indirect generation of ROS in target cells exposed to a variety of engineered or ambient NPs is a major mechanism leading to cell toxicity and cell death (Nel et al., 2006; Unfried et al., 2007). Excess ROS that are not detoxified by endogenous antioxidant defenses can damage cellular membranes via lipid peroxidation, inactivate structural proteins, enzymes, and ion pumps, and damage nuclear DNA. A “hierarchical stress response” to oxidant-induced injury has been proposed by Nel et al. (2006). Low levels of ROS can activate transcription factors leading to upregulated expression of antioxidant defense-related genes. The second tier of response to ROS is activation of redox-sensitive transcription factors that upregulate genes involved in inflammation, proinflammatory cytokines and chemokines (Albrecht et al., 2004). Finally, high or sustained production of ROS can lead to irreversible mitochondrial injury and cell death by apoptosis or necrosis (Nel et al., 2006).

Endocytosis and phagocytosis of NPs by target cells usually results in fusion with and sequestration in membrane-bound lysosomes (Russell et al., 2009; Chitharani, 2010). Uptake of quantum dots, CNTs containing metal catalyst residues, or metallic and metal oxide NPs provides a pathway for intracellular delivery of toxic metal ions by a “Trojan horse mechanism” (Limbach et al., 2005). In the acidic environment of lysosomes, cadmium ions can be released from quantum dots, redox-active iron from iron oxide NPs, and other toxic metal ions including nickel from CNTs or metallic and metal oxide NPs (Liu et al., 2008; Pietruska et al., 2011). This mechanism may also be exploited to release drug cargoes into the cytoplasm following NP-mediated delivery (De Jong and Borm, 2008).

Cellular uptake of long, rigid, high aspect ratio nanomaterials including CNTs, metallic nanowires, and nanorods is especially hazardous for target cells in the lungs (Donaldson et al., 2010). Macrophages are the initial cells to phagocytize inhaled particulates deposited in the airways or alveoli (Geiser, 2010). Rigid, high aspect ratio ENMs show similar interactions with macrophages as amphibole asbestos fibers. If they are longer than the diameter of macrophages (~10 μm), rigid elongated nanostructures show incomplete uptake or phagocytosis with prolonged generation of ROS by the respiratory burst mechanism of phagocytes and extracellular release of damaging lysosomal enzymes. Rigid, high aspect ratio nanomaterials also cause permeabilization of the lysosomal membrane resulting in release of cathepsins into the cytoplasm and activation of the inflammasome (Hamilton et al., 2009). Cathepsins are proteases active at neutral pH and can cleave precursors of proinflammatory cytokines that can initiate inflammatory reactions in the lungs (Biswas et al., 2011) as well as activate procaspases leading to death by apoptosis (Boya and Kroemer, 2008). Activation of the inflammasome, the cytoplasmic protein complex where this proteolytic cleavage and activation occurs, is also triggered by uptake of crystalline minerals including silica and asbestos fibers (Dostert et al., 2008).

Incomplete sequestration of long, rigid, high aspect ratio NPs in lysosomes may also promote their release into the cytoplasm resulting in physical interference with cytoskeletal function. Cytoskeletal disruption can cause impaired cell motility, for example, macrophage-mediated clearance of asbestos fibers and high aspect ratio NPs is impaired resulting in sustained lung injury, inflammation, and fibrosis (Donaldson et al., 2010) as discussed below. In other organs, NP-mediated interference with cytoskeletal-mediated transport may occur, for example, impaired bile transport and secretion by hepatocytes (Johnston et al., 2010a). In dividing cells, high aspect ratio NPs may physically interfere with the mitotic apparatus resulting in chromosomal missegregation and polyploidy (Sargent et al., 2010).

Finally, NPs may translocate into the nucleus by diffusion through nuclear pores or via nucleocytoplasmic transport (Unfried et al., 2007). Quantum dots (Hoshino et al., 2004), silicon dioxide NPs (Chen and von Mikecz, 2005), and gold NPs (Pante and Kunn, 2002) have been detected in the nucleus. Small gold NPs may intercalate into DNA (Schmid, 2008) and generate ROS that may induce oxidative DNA damage (Kang et al., 2010). Silicon dioxide NPs have been shown to induce intranuclear protein aggregates of histones, topoisomerase I, and fibrillarin, although the functional consequences of this intranuclear protein aggregation are unknown (Unfried et al., 2007).

The unique physical and chemical properties of solid materials at the nanoscale contribute to serious technical limitations and potentially misleading results using conventional toxicology assays. Due to their high surface area and hydrophobicity, NPs can adsorb vital dyes, cell culture micronutrients, or released cytokines (Monteiro-Riviere and Inman, 2006; Casey et al., 2007,2008; Guo et al., 2008; Stone et al., 2009). High surface area carbon nanomaterials such as SWCNTs and graphene have the greatest potential to adsorb aromatic amino acids and vitamins by π–π interactions with their hydrophobic graphenic surfaces (Guo et al., 2008). Adsorption of bacterial lipopolysaccharide or endotoxin (a component of the cell wall of gram-negative bacteria) is a major concern because even low levels of endotoxin can activate macrophages and dendritic cells (Vallhov et al., 2006). Genotoxicity endpoints are especially sensitive to NP-induced artifacts, for example, particle agglomerates interfere with the comet assay, used to detect DNA breaks and cytochalasin B used in the cytokinesis-arrested micronucleus assay, can interfere with NP uptake (Doak et al., 2009; Stone et al., 2009). NP agglomerates in physiological saline may also induce greater cytotoxicity than well-dispersed NPs; however, controls must be included to rule out toxicity of the dispersant (Wick et al., 2007). In all toxicological studies, well-characterized positive and negative reference samples should be included over a range of doses and exposure times.

ENMs are highly complex and have unique chemical and physical properties relative to bulk materials of the same chemical composition. Complete materials characterization is required for adequate interpretation of toxicological studies (Warheit, 2008; Stone et al., 2009). Nanotoxicologists must collaborate with chemists, engineers, and materials scientists in an interdisciplinary research team in order to understand the chemical, physical, and structural properties responsible for biological interactions and potential toxicity of nanomaterials. The relevance of in vivo extrapolation from high-dose, short-term, in vitro toxicity assays is discussed in more detail below.

In principle, it should be possible to engineer NPs with desirable surface properties for commercial or biomedical applications, for example, to control their size and geometry in order to prevent indiscriminate entry into cells and frustrated phagocytosis. Capping or coating of NPs using antioxidants such as tocopheryl-polyethylene-glycol-succinate (Yan et al., 2007) may decrease toxicity. Release of toxic metal ions from quantum dots and iron oxide NPs can be minimized using inorganic shells or biocompatible polymers such as chitosan or alginate (De Jong and Borm, 2008). CNTs, fullerenes, and graphene are easily surface functionalized; for example, covalent functionalization with carboxyl groups has been reported to decrease toxicity of these pristine carbon nanomaterials (Sayes et al., 2005, 2006; Sasidharan et al., 2011). There is some evidence that CNTs are less pathogenic if short in length, or if entangled to hide their fibrous nature (Poland et al., 2008). A long-term goal of mechanistic nanotoxicology is to reveal structure-activity relations that may allow the design of safe nanomaterials through re-engineering or reformulation (Hutchison et al., 2008).

Zinc oxide and TiO₂ NPs are widely used in sunscreens because they show less scattering of visible light and appear transparent, while effectively absorbing UV light in both the UVB and UVA wavelengths (Wang and Tooley, 2011). Due to its rapid dissolution in water releasing toxic Zn²⁺ ions, zinc oxide NPs are considered as potential environmental toxicants (Kahru and Dubourguier, 2010). Xia et al. (2011) deliberately doped zinc oxide NPs with iron (1%–10% by weight) that significantly retarded release of Zn²⁺ ions in water and physiological saline solutions. Zn²⁺ ions inhibit hatching of zebrafish embryos. Pristine zinc oxide NPs also inhibited hatching; however, iron-doped NPs were less inhibitory in this assay. Similarly, iron-doped zinc NPs induced significantly less lung injury and inflammation than pristine zinc oxide NPs following intratracheal instillation in rodents. This study provides proof-of-principle for safe design of ENPs to minimize potential environmental toxicity and adverse health effects.

TiO₂ NPs are used commercially in foods (candy, chewing gum), personal care products (shampoos, shaving creams, toothpastes), sunscreens, and paints and photocatalytic coatings for buildings and windows (Weir et al., 2012). In 2010, 5000 metric tons of TiO₂ NPs were produced (Landsiedel et al., 2010) and have been detected in effluents from waste water treatment plants where they may enter surface water (Westerhoff et al., 2011).

The potential of zinc oxide and TiO₂ NPs to induce phototoxicity and penetrate into the dermis has been a major concern for human safety of sunscreens (Wang and Tooley, 2011). These mineral-based sunscreens cause less skin irritation and produce fewer allergic reactions than chemical-based sunscreens. The pristine NPs used in sunscreens form aggregates 30 to 150 nm in diameter; in this size range, these NPs are photoreactive and may generate ROS. Although uncoated TiO₂ NPs exhibit phototoxicity, following coating with hydrophobic stabilizers used in sunscreen formulations, no phototoxicity was detected in a cell culture assay system (Schilling et al., 2010). A series of skin penetration studies using both ex vivo and in vivo models also showed that these NPs do not penetrate deeper than the outer most layer or stratum corneum of intact skin (Schilling et al., 2010). However, TiO₂ NPs suspended in cyclopentasiloxane fluid used in cosmetics did penetrate into hair follicles of shaved skin of micropigs (Senzui et al., 2010). Finally, young pigs were exposed to UVB light producing sunburn and an in vitro skin perfusion system was used to assess penetration of sunscreens containing zinc oxide or TiO₂ NPs. Only superficial penetration into the upper layers of the epidermis was observed and no zinc or titanium above untreated control levels was observed in the perfusate (Monteiro-Riviere et al., 2011). Collectively, these in vitro and in vivo toxicology assays suggest that the benefits of protection against carcinogenic UV light radiation provided by sunscreens formulated with zinc oxide or TiO₂ NPs outweigh the minimal risks associated with phototoxicity, DNA damage, and skin penetration into shaved or sunburned skin.

Introduction

Nanotoxicology, the science of effects of ENMs on living organisms, is a rapidly growing discipline aimed at identifying and characterizing nanomaterial toxicity that will serve—in combination with exposure data—the ultimate goal of performing a meaningful risk assessment (National Academy of Sciences, 1983). Many ENMs exhibit unique and desirable catalytic, optical, structural, or electronic properties that make them attractive in diverse technological areas, including multiple manufacturing industries and environmental and medical applications. At the same time, concerns have been raised regarding potential acute and chronic adverse effects due to their physicochemical properties in combination with nanostructures, for example, an elongated fiber-like shape of CNTs (Johnston et al., 2010b).

CNTs are a prime example of the two opposing faces of nanomaterials: Many highly desirable properties that are suitable for numerous beneficial applications contrast with reports of serious adverse effects in experimental animals. For example, the excitement about future uses of CNTs in biomedical applications for delivery of drugs, genes, and biosensors or as tissue engineering scaffolds (Liu et al., 2008; Li et al., 2010; Liang and Chen, 2010) is dampened by reports of inflammatory fibrogenic and even mesotheliogenic effects in laboratory rodents (Shvedova et al., 2005; Takagi et al., 2008; Jacobsen et al., 2008; Poland et al., 2008; Sakamoto et al., 2009). Results of a number of in vitro studies in different cell systems also revealed an oxidative stress-inducing proinflammatory potential of CNTs (Hirano et al., 2010; Vankoningsloo et al., 2010; Pacurari et al., 2008; Sanchez et al., 2011; Monteiro-Riviere et al., 2005); however, other in vitro studies did not confirm such effects (De Nicola et al., 2007; Fenoglio et al., 2006; Flahaut et al., 2006; Pulskamp et al., 2007).

As will be discussed later, a common problem with in vitro studies is that very high doses/concentrations in culture media are used that have no relevance for realistic in vivo conditions. Moreover, results of in vitro assays can be misleading due to interference of the nanomaterials to be tested with the testing reagents, adsorption of induced mediators (Worle-Knirsch et al., 2006; Han et al., 2011), or interference with optical measurements (Pfaller et al., 2009); general problems related to sensitivity and reliability of in vitro systems have also been discussed (Jones and Grainger, 2009). A most recent example of reliability problems of in vitro assays in evaluating the toxicity of CNTs is described by Haniu et al. (2011) reporting contradictory results of increased as well as decreased cell viability as well as either no or high cytokine release when the same CNT and the same cell type were dosed using different dispersants for pretreatment.

Concepts of Nanotoxicology

The previous section discussed the importance of physicochemical properties of ENM. Table 28-3 summarizes several of these properties that are known to affect biological/toxicological effects. The size and size distribution are important determinants—thermodynamic and aerodynamic—for the deposition efficiency of inhaled materials throughout the respiratory tract. (See Fig. 28-6.) The quality of dispersion in liquid media can be determined by measuring the hydrodynamic diameter, which is important information for in vitro studies and the ENM behavior in blood and lymph circulation. The shape of ENM (tubes, fibers, planes) influences their aerodynamic and hydrodynamic behavior and is of special significance for cellular uptake, for example, a fiber shape of straight nanotubes with a length exceeding the diameter of alveolar macrophages so that they cannot be completely phagocytosed (frustrated or incomplete phagocytosis) resulting in potent activation and release of inflammatory mediators (Kane and Hurt, 2008). Individual NPs tend to agglomerate (held together by relatively weak forces, eg, Van der Waals forces) or they are aggregated when produced (sintered together or firmly bundled in the case of fibers) (Teleki et al., 2008; ISO/TS 27687, 2008; Taurozzi et al., 2011). Oftentimes, larger structures in the form of agglomerated aggregates are formed (Fig. 28-6), both in the airborne state and as liquid suspensions.

The terms “agglomeration” and “aggregation” are often used interchangeably, yet ISO/DIS (2011) defines them as based on the forces that keep the primary particles together (agglomerates: weakly bound with external surface area similar to the sum of surface areas of individual components; aggregates: strongly bonded or fused with the external surface significantly smaller than the sum of calculated surface areas of individual components). Both agglomerates and aggregates are also termed “secondary particles” (ISO/DIS, 2011).

The formation of agglomerates of aerosols (coagulation) depends on their number concentration in the airborne state and is a function of time (Hinds, 1982). The effect of such time-dependent agglomeration on decreasing the number of airborne particle concentration and on increasing the overall particle size is illustrated in Table 28-5 assuming simple monodisperse coagulation in still, undisturbed air. For example, it takes only 20 μs for a very high number concentration of 10¹⁴ particles/cm³ to decrease by half, whereas it takes 231 days if the initial concentration is only 100 particles/cm³. Similarly, in order to double the initial particle size because of a resulting increase in the agglomerate particle size, it takes 140 μs for the initially very high concentration, but four years for the low concentration. This dramatic shift in particle size distribution needs to be considered when high concentrations of airborne NPs are measured close to their source.

The agglomeration and the aggregation state of ENM will significantly influence their bulk and apparent density. For example, the specific density of TiO₂ is ~4 g/cm³, whereas the bulk density (density of a powdered material including interparticle spaces) of nano-TiO₂ depending on primary particle size ranges from 0.13 to 0.54 g/cm³. Bulk density—interchangeably also described as apparent or packing density—can be measured by pouring the ENM powder into a fixed volume of a graduated cylinder (freely settled). The volume of the measured mass/cm³ includes the particle volume, the interparticle void volume, and any internal pore volume. Using MWCNTs as an example, the specific density of the base material, carbon, is ~2 g/cm³, but the bulk density of MWCNT was reported as ranging between 0.043 and 0.31 g/cm³ (Ma-Hock et al., 2009; Pauluhn, 2010). The density achieved by packing of individual agglomerate structures of different MWCNT obviously plays an important role for differences of the bulk or packing density, as do the dimensions of length and diameter, the straightness of the tubes and the form and size of tangled structures. Knowledge of bulk density is important for understanding the aerodynamic behavior of ENM and their deposition in the respiratory tract. In simple general terms, the aerodynamic diameter (D_ae) of an aerosol is correlated with the physical geometric particle diameter (D_geom) as D_ae = D_geom × Display Formula$rho$ (rho = material density in g/cm³). For particles below 1 μm, a slip-correction factor needs to be added. For highly agglomerated ENM, it may be preferred to replace the true material density with the value of bulk or apparent density in order to more accurately estimate the aerodynamic diameter.

A low bulk density obviously increases the volume of ENM after uptake into cells relative to the volume occupied by the same mass of nonagglomerated material. A volumetric overload concept for alveolar macrophages has been proposed by Morrow (1988), who suggested that when a phagocytized volume of 60% of the normal macrophage volume is reached, movement and clearance function is completely abolished, and that a retardation of the clearance function of macrophages starts when ~6% of the cell’s volume is exceeded by phagocytized particles. Pauluhn (2010) applied this concept of volumetric overload to results of his subchronic inhalation study with MWCNT in rats for estimating an occupational exposure level.

Surface characteristics of ENM are a critical determinant for interpreting dose–response relationships (surface area as dosemetric), and for nanomaterial cell interactions. Uptake into cells is influenced by their surface charge, surface reactivity, the chemistry of surface coatings [functional groups, polymeric coatings] and also surface defects to the material as-synthesized, or introduced during surface functionalization or processing such as acid purification or ultrasonication for dispersion).

Solubility and solubilization rate are of interest because the shortened biopersistence of NP in the organism terminates any NP-size-related effects. Zinc oxide is an example of an NP that undergoes rapid dissolution, thereby giving rise to high ionic zinc concentrations and induction of inflammation and oxidative stress (Xia et al., 2008b) in contrast to poorly soluble CeO₂ or TiO₂ NPs. Many ENPs are insoluble in the as-produced form (eg, metallic Ag⁺ and Ni²⁺ metallic NPs), and do not undergo simple dissolution, but can undergo chemical oxidation, a kind of nano-corrosion, in solution, in tissue, or in the environment to produce soluble species (eg, Ag⁺ and Ni²⁺) in a process that gradually degrades and eliminates the particle state (Liu and Hurt, 2010; Liu et al., 2010a; Pietruska et al., 2011). Such NPs can act via a “Trojan Horse” mechanism in that they are taken up into cells and subsequently dissolve, thereby creating a very high intracellular ionic metal concentration that is cytotoxic (Limbach et al, 2007). The same ionic concentration outside the cell will elicit no or only a modest response because of selective cell barriers preventing metal ions from entering the cell. The results by Xia et al. (2008b) and Limbach et al. (2007) are based on in vitro studies. Donaldson et al. (2009) have cautioned that great care should be taken in interpreting that oxidative stress induction by particles observed in vitro will also occur in vivo. This is not to say that the “Trojan Horse” effect does not occur in vivo, it has been described and is being used in designing drug delivery via NP for therapeutic applications (Iezzi et al., 2012; Kreuter, 2004). The cautionary note is more generally directed toward the extrapolation of in vitro-based oxidative stress results to in vivo conditions.

It is a major challenge for nanotoxicological studies that many of these properties, once they have been determined for a given ENM, are not stable but can change when prepared for toxicological testing or during storage or testing. Alterations can occur through oxidation or dissolution (above), or through adsorption of proteins and lipids once they have been taken up into different compartments of the organism, has become an important area of research (Cederval et al., 2007a,b; Lundqvist et al., 2008; Monopoli et al., 2011), yet methods to study these changes on the surface of ENMs in vivo still need to be developed.

Dosemetrics

Given the importance of physicochemical properties of ENM as determinants of their biological/toxicological properties, the question of the proper dosemetrics has been raised. Toxicologists are used to thinking in units of mass or molarity when characterizing a dose or concentration. However, several in vitro and in vivo studies with NP have shown that expressing dose–response relationships on a convenient mass basis may not be as informative as the surface area of the NP (Oberdörster et al., 1992; Donaldson et al., 1996; Tran et al., 2000). Fig. 28-8 shows the pulmonary inflammatory dose–response relationships of two sizes of TiO₂ particles induced by intratracheal instillation in rats. A significantly greater influx of inflammatory neutrophils into the lung was induced by 25 nm TiO₂ per unit mass than by 250 nm TiO₂. The result is a very steep dose–response for the nanosized TiO₂ and a flatter dose–response for the larger TiO₂. Likewise, there was a clear separation of the dose–response when based on the number as dosemetric; however, when the same data were expressed based on particle surface area, a common dose–response relationship emerged (Fig. 28-8).

Although the concept of particle surface area as dosemetric is biologically plausible because it is the surface of particles that interacts with cell structures, the total surface area may only be a surrogate of the actual biologically relevant surface. As was discussed above, there are a number of surface properties (Table 28-3) that affect interaction with cells. Thus, identifying a biologically available surface area will be of great value for defining a proper dosemetric. At present, the specific surface area of ENMs (m²/g) is most commonly measured by vapor adsorption, in which the number of vapor molecules (usually nitrogen) adsorbed on solid surfaces is measured over a range of standard conditions and the surface area estimated by application of the BET theory (Brunauer et al., 1938). Both the external and existing inner surface (porosity) in pores larger than the nitrogen molecule are determined by this technique. There is a need for standardization of the method with regard to the use of nitrogen or other gases (Kr; Ar) because different molecule sizes will “see” and report different surface areas in molecular-sized pores (<0.4 nm). Measurements of BET surface area can also contribute to differentiation between agglomerated and aggregated structures, for example, agglomerated NPs have essentially the same specific surface area as the same mass of individual primary particles, whereas aggregated particles have a somewhat lower specific surface area than the same mass of primary particles (ISO/TS 27687, 2008). The measured surface area can then be converted to a BET equivalent particle size for nonporous spherical particles (Teleki et al., 2008).

In addition to mass, number and surface area as dimensions for a dosemetric, the volume of NPs has been suggested as another dosemetric. This suggestion is based on the interpretation of earlier long-term inhalation studies in rats in which very high concentrations of different poorly soluble particles of low cytotoxicity had resulted in highly overloaded alveolar macrophages so that their normal clearance function was severely impaired or had come to a standstill. Based on the results of these studies, Morrow (1988) formulated a “particle overload” hypothesis: When the volume of phagocytized particles in alveolar macrophages exceeds 6% of the normal macrophage volume, their physiological clearance function becomes impaired; if the phagocytized volume reaches 60%, the clearance function ceases completely. For certain ENMs the volume that they occupy can be very high, for example, larger agglomerates of MWCNT, forming a loosely packed (low specific density) structure that takes up a large volume when taken up by alveolar macrophages in the lung. Pauluhn (2010) has applied this volumetric overload concept to results of a well-performed subchronic rat inhalation study for estimation of a human occupational exposure limit.

Follow-up studies to the mechanistic basis of lung particle overload point to particle surface area of agglomerated phagocytized nanosized and larger particles as dosemetric, which appears to correlate better with the impaired clearance function of alveolar macrophages rather than the phagocytized volume (Oberdörster et al., 1994; Tran et al., 2000). Whether particle volume or particle surface area are the more appropriate dosemetric to use in situations of particle overload of the lung is still to be determined, yet—as pointed out before—it is important to remember that not only physical –volume, surface area—but also chemical—surface associated—properties of NPs are critical determinants of effects resulting from NP–cell interactions (Table 28-3).

Among the different dosemetrics discussed in this section, particle number is of particular importance with respect to characterizing exposure. Exposure concentrations in the air are expected to be very low, which makes it difficult to determine the precise concentration based on measuring mass. However, a low mass of NP consists of a huge number of particles that can reliably be measured in the airborne state using scanning mobility particle sizers or concentration particle counters. Thus, NP number as an exposure-metric for mainly isometric type NP is a valuable component for exposure characterization.

Table 28-6 shows the correlation among the different dosemetrics of different 50 nm-sized isometric NPs, with specific material densities ranging from 21 to 1 g/cm³. Volume is not listed in this table because it depends on the bulk or packing density, which is different from the material density because of additional void spaces between the packed NP. Surface area and number per unit mass and per 100 μg/m³ air volume vary greatly. The table also shows that a 200-fold change of particle size from 5 nm to 1000 nm results in a more than six orders of magnitude change in number concentration with only a 20-fold change in surface area for all NP (shown for polystyrene in Table 28-6). However, although these correlations can readily be established for dry powders, even when they consist of polydisperse particle sizes, these correlations will change upon inhalation and disposition in the body because of different deposition and biodistribution efficiencies in the lungs and other organs.

Portals of Entry

The respiratory tract, the gastrointestinal (GI) tract, and skin are the main organs of direct exposure to ENM. For medical application, injection (intravenous, intramuscular, subcutaneous, and other) will also be an important entry route into the organism. Intake via the respiratory tract is a most prevalent exposure route for occupational exposures; additives of ENM to food (TiO₂; SiO₂) and potential contamination of food from nano-enabled packaging materials result in exposure via GI tract (Jani et al., 1990; 1994; Hoet et al., 2004; FAO/WHO, 2009; Loeschner et al., 2011). Based on available data, translocation of nanomaterials in vivo across GI-tract epithelial cells seems to be limited; Trouiller et al. (2009), however, described DNA damage in bone marrow cells following very high gavage dosing of rats. Skin exposure via cosmetics and skin-care products occurs, although penetration of healthy skin by NP has not been demonstrated (Monteiro-Riviere et al., 2011; Prow et al., 2011; Schneider et al., 2009). Some penetration of quantum dots (~30 nm) applied to the skin of mice after UV radiation-induced sunburn was found (Mortensen et al., 2008). Because the respiratory tract is a major route for humans to exposure of nanomaterials, the following sections will focus on this portal of entry to discuss biokinetics, effects, and concepts for in vivo and in vitro toxicity testing.

Dosing of the Respiratory Tract

The following sections focus on the respiratory tract as portal-of-entry of airborne ENM.

Dosing of the respiratory tract of laboratory rodents as the portal-of-entry for airborne ENMs is mostly and conveniently done by intratracheal instillation (rats) or oropharyngeal aspiration and intranasal instillation (mice), which administer materials as a bolus in a second or less. However, inhalation is the only physiological method and should be considered the gold standard for exposure to airborne materials. Obvious major differences between bolus-type and inhalation exposures relate to the dose rate (see discussion later), use of anesthesia, the distribution of administered material within the respiratory tract, and the need for special expertise and equipment required for inhalation but not bolus-type exposures. Although most of the differences between inhalation- and instillation-type delivery are well known and have been discussed in a Society of Toxicology (SOT)-sponsored White Paper (Driscoll et al., 2000), the limitations of using and interpreting results from bolus-type dosing are most often not considered (Table 28-7). Driscoll et al. concluded with several recommendations when intratracheal instillation can be applied. These included evaluation of comparative toxicity: screening, ranking, range of doses, comparing to reference material; evaluation of material not respirable by rodents; to minimize interference of clumping and localized inflammation, only <~100 μg/rat should be used. The authors also mention situations when instillation should not be considered: For determining deposition pattern in lung to mimic inhalation; toxic effects in the upper respiratory tract; not to use for materials that react with vehicle; they also cautioned that short-term clearance (mucociliary) is not reflected accurately.

Of special importance is the tremendous difference in the delivered dose rate: Bolus-type delivery occurs within a fraction of a second, whereas inhalation at realistic concentrations takes hours to months of exposure in order to deposit the same dose in the lung. Treating a dose delivered by bolus to be the same as a dose that has accumulated in the lung over a life-long exposure is scientifically not justifiable and needs to be discouraged (Oberdörster, 2012). Inundating cells abruptly with an extraordinarily high dose does overwhelm the cell’s defense mechanisms and leaves no time for developing adaptive responses. Consequently, underlying mechanisms of effects induced by such unrealistic high doses are different from those induced by relevant doses and dose rates (Slikker et al., 2004, see Fig. 28-3). With respect to the administered dose rate, Fig. 28-9 illustrates a tremendous difference of inducing a pulmonary inflammation by either intratracheally instilling (~0.5 second duration) 200 μg of TiO₂ NP versus depositing the same dose by inhalation over a period of four hours or four days (4 × 4 hours). The difference in dose rate is four to five orders of magnitude, with a strong inflammatory response at the highest dose rate (instillation) and no response at the lowest dose rate (inhalation).

Evidence that extremely high doses acutely administered within a very short time overwhelm the organism’s or the cell’s defenses so an adaptive response cannot be mounted was presented in a rat inhalation study with nanosized particles using exposure to UFPs (~18 nm count median diameter) of polytetrafluoroethylene (PTFE) fumes (Johnston et al., 2000). PTFE fumes, similar to metal fumes, when freshly generated consist of nanosized particles, which are highly reactive and are known to give rise to polymer fume and metal fume fever in occupationally exposed workers (Drinker et al., 1927a,b; Drinker and Drinker,1928; Seidel et al., 1991; Lee and Seidel, 1991). When groups of rats were exposed for five minutes on each of three days to either filtered air or to a concentration (~50 μm³) of these PTFE fumes, followed on day four by a 15-minute exposure of both groups to the same concentration of PTFE fumes, the pre-exposed rats survived without any significant adverse pulmonary responses (not significantly different from rats that were sham exposed on all days), whereas the filtered air pre-exposed rats died within three hours after the 15-minute PTFE exposure (Fig. 28-10a). Estimated deposited dose of PTFE in the lung was only ~60 ng. Further lung tissue analyses revealed that the PTFE pre-exposed rats had mounted a strong “adaptive” response (increased antioxidant and chemokine mRNA), whereas nonadapted rats were unable to induce such response (Fig. 28-10b). Evidence for such adaptation following inhalation exposure to nanosized particles in humans was already reported eight decades ago when Drinker et al. (1927b) described that repeat exposure to zinc oxide fumes induced “resistance” in human subjects. Adaptive responses are important physiological protective mechanisms, which need to be considered when interpreting results of nanomaterial toxicity testing. However, the organism’s capacity to adapt may not only be compromised by sudden very high acute exposures of short duration (high dose and high dose rate) but may also be impaired in susceptible parts of the population.

Despite the limitations of bolus-type delivery, they may be viewed as “proof of principle” or “hypothesis-forming studies,” with the findings to be confirmed by subsequent inhalation studies. Although results from bolus-type studies are useful for toxicity ranking against known positive or negative control ENMs, they cannot be used for quantitative risk assessment. Other in vivo bolus-type deliveries involve intravenous, intraperitoneal, or intrapleural injections. If used for safety assessment of medicinal applications of ENMs, including CNTs, an appropriate range of doses should be applied. If, on the other hand, intravenous or intracavitary injections are used to simulate doses resulting from translocation of inhaled CNTs after deposition in the lung, results have to be viewed with caution. The amount of NPs translocating from the lung to the blood circulation and subsequently to secondary organs are very low, on the order of 1% to 3% (Kreyling et al., 2002; 2009). Furthermore, the translocation rate from the lung is very low, it occurs over hours and days, which contrasts with the immediate very high dose rate of injection. Bolus-associated problems have been discussed above, and confirmation of results from such “proof of principle” studies by follow-up inhalation studies is desirable.

The pretreatment of ENMs in preparation of bolus-delivery dosing is an additional difference compared to inhalation and may raise some concern: The recommendation to use dispersants in order to administer the materials in a well-dispersed mode (Fubini et al., 2010; Taurozzi et al., 2011) results in the coating of the treated nanomaterials that changes their surface properties and renders them different relative to their pristine state when inhaled under real-world exposure conditions. Although inhaled ENMs will also be coated by lipids and proteins of the lung lining fluid once deposited, such protein corona is likely to change upon translocation to interstitial and blood compartments based on the “concept of differential adsorption” (Müller and Heinemann, 1989). This concept states that the physicochemical properties of nanomaterials, such as size, surface properties, shape, dissolution, and others when in contact with media in the different body compartments determine protein and lipid adsorption and desorption patterns and thereby influence biodistribution across barriers and in target tissues and cells (Fig. 28-11). This concept has been further developed more recently by Dawson’s group (Cederval et al., 2007a; Lundqvist et al., 2008; Walcyzk et al., 2010) who describe the formation of hard and soft protein coronas around NPs upon incubation with blood plasma. Future studies have to show how pretreatment of ENMs for bolus-type in vivo administration will affect both biodistribution and effects. In vitro experiments with different cell types have already demonstrated that the use of different dispersants with MWCNTs resulted in contradictory results (Haniu et al., 2011).

The three different methods of bolus-type dosing of the respiratory tract—intranasal instillation; intratacheal instillation; oropharyngeal aspiration—represent a convenient way to obtain first or preliminary data about effects in the respiratory tract. The limitations and caveats of using these modes of administration—as pointed out in the preceding paragraphs—should be taken into account when interpreting the results. Immediate short-term inflammatory responses will be induced, and these early bolus-associated effects are likely to also influence long-term effects. For example, when Li et al. (2007) compared pulmonary responses to inhaled and intratracheally instilled MWCNTs in rats, they observed a much lower response in terms of pulmonary pathology after inhalation despite a higher lung burden from inhalation compared to instillation. It is of interest though, that in another study similar inflammatory, granulomatous and fibrotic responses were observed in mice following inhalation and oropharyngeal aspiration of SWCNTs. In fact, inhalation exposure was more potent than aspiration of an equivalent mass (estimated) of SWCNTs one to 28 days after inhalation exposure to a high concentration (5 mg/m³, four days @ 5 h/day) or oropharyngeal aspiration of 5 and 10 μg (Shvedova et al., 2008). The authors suggested that the similarity of responses between inhalation and aspiration exposure to SWCNTs in mice may be due to smaller SWCNT structures by inhalation of a dry aerosol versus aspiration of a suspension containing micrometer-sized agglomerates. However, both the Li et al. (2007) and Shvedova et al. (2008) studies only estimated the doses deposited in the lung upon inhalation based on inhaled concentration and size distribution and retained doses were not confirmed by actual measurements. In addition, the inhalation exposure by Li et al. was carried out with a static system, that is, MWCNTs were introduced into the exposure chamber as an aerosol and kept there for 90 minutes to slowly settle down while the rats were exposed. This makes it very difficult to determine the actual exposure concentration because it continually decreased, and changes in particle size due to rapid settling of the aerodynamically larger particles will occur. A dynamic exposure system with continuously generated aerosol is the standard and preferred method for inhalation exposures.

Respiratory Tract Deposition

Inhalation by humans of airborne ENM at the workplace or when handling powders of ENM results in significant deposition in the three compartments of the respiratory tract: the nasopharyngeal region from nose/mouth to the larynx; the tracheobronchial region from larynx to terminal bronchioles; and the alveolar region from the first generation of respiratory bronchioles (bronchioles with some alveoli) to the last generation of alveolar ducts. The deposition efficiency of inhaled particles depends on several particle characteristics, the anatomical structure of the airways, and breathing parameters. Particle size, size distribution, density, and shape are most important for their aerodynamic and thermodynamic diameter, which govern deposition in the respiratory tract by inertial impact, gravitational settling, and displacement by diffusion. In addition, interception (in particular for elongated structures or fibers) and electrostatic image forces (positive charge on particles attracted by negative epithelial surface) play a role in certain conditions (see Chap. 15).

Because in vivo toxicological studies are mostly performed in experimental animals, and because of anatomical and some physiological differences in the respiratory system, knowledge about variations in the behavior of inhaled particles is essential for interpreting results of animal inhalation studies with respect to dosimetric extrapolation to humans. Several particle deposition models have been developed based on mathematical modeling and supported by results of numerous human studies using inhalation of benign particles of different sizes. The two most frequently used models are the extensive international commission on radiation protection (ICRP) (1994) model for humans, and a user-friendly (computer) Multiple Path Particle Dosimetry (MPPD) model for rats and humans (Asgharian et al., 1999). Each of these models is based on results of an extensive database of particle deposition studies in humans (ICR Panel MPPD) and in rats (MPPD). These data have been incorporated into mathematical equations describing airborne particle behavior due to impaction, sedimentation, and diffusion for different airway geometrics in humans and rats. There is good general agreement between results of the two models, although some quantitative differences exist. It should be kept in mind that these are still models to estimate the deposition in the respiratory tract depending on particle parameters, airway geometrics, and breathing modes. Interindividual time values can be very different from the averages that are calculated by the models.

Fig. 28-12 compares the deposition fractions for monodispersed particles of sizes from about 1 nm to 20 μm diameter between humans and rats based on the MPPD model, assuming unit density material. For both species nasal breathing is assumed, which is the obligatory mode for rats. For NPs below 100 nm (in the thermodynamic size range, less than ~0.5 μm), there is only a small difference in deposition between nasal- and oral-breathing mode, whereas these differences become more pronounced for larger particles in the aerodynamic size range (starting at ~0.5 μm). The mechanisms by which inhaled particles are depositing changes from diffusion (due to bombardment of particles by air molecules) as exclusive mechanism for airborne single NPs to gravitational sedimentation and inertial impaction for the larger particles. Comparing deposition efficiencies between rats and humans, similarities but also differences become obvious: Qualitatively, size-dependent deposition fractions are similar; for example, there is a lowest point of deposition for inhaled particles in both species around the size of 0.5 μm, which is due to the fact that at this size the combined impact of diffusional displacement and gravitational settling is minimal, so they are most persistent in the air. Of interest is that—contrary to a general misconception—the smallest inhaled particles (<5 nm) are not depositing in the most distal areas of the lung, but in the upper extrathoracic region. Thus, the nasal filtering capacity is effective for both the smallest and largest particles.

Obvious differences between rats and humans are the maximum size of particles that are respirable, that is, will reach the alveolar region. In rats, this is about 5 μm aerodynamic size, in humans about 15 μm. Although these sizes are outside the range of single NP, airborne NP occur for the most part as agglomerates, which can reach μm size equivalent aerodynamic diameters. The concept of rat versus human respirability also applies to smaller μm size particles. Even for submicron-sized particles rat/human deposition fractions are not necessarily the same. Thus, if a rat inhalation study is designed to simulate a human exposure scenario with an aerodynamic median particle size of 5 μm, not much of the aerosol will reach the rat’s alveolar region. Instead, attempts should be made to streamline the particle size for the rat study such that the deposited dose in the alveolar region is similar between the two species relative to alveolar surface area.

The deposition curves in Fig. 28-12 imply that inhaled particles of ~20 nm have similar deposition efficiencies in all three regions of the respiratory tract. However, a more judicious analysis reveals that the alveolar region actually receives the least amount of these NPs when the deposited dose is expressed per unit surface area (Fig. 28-13). The data in this figure are derived from the MPPD model, assuming an eight-hour exposure to 20 nm monodispersed TiO₂ NPs under nasal-breathing conditions for rats and humans. As can be seen, the deposited dose normalized for surface area is less than 1 ng/cm² in the alveolar region, and is highest for the extrathoracic nasopharyngeal region. This result should also serve as a reminder that realistic in vivo doses to cells of the respiratory tract are mostly orders of magnitude lower than doses that are typically applied in vitro to lung epithelial cell cultures. Moreover, the deposited dose in Fig. 28-13 had accumulated over an eight-hour exposure, which is different from a high in vitro dose rate, similar to dose rate differences between in vivo instillation and inhalation (see Fig. 28-8).

The development of deposition hotspots in the respiratory tract during particle inhalation is often cited as justification for using high doses in vitro studies (Phalen et al., 2006). These hotspots occur at the bifurcations (carinal ridges) of the upper generations of the tracheobronchial region where due to impaction more particles per unit surface will deposit compared to the average deposition per unit surface over the length of the airways (Fig. 28-14). Concentrations up to several 100-fold higher than on average in the airway may be accumulated for larger particles. However, as Balásházy et al. (2003) pointed out using computer modeling, the size of the area receiving the highest concentration is the size of square (patch) with a sidelength of only 0.1 nm, and airborne NPs are contributing to these hotspots to a lesser degree only: for a patch size of 1 mm the enhancement factor is about 30-fold for 1 to 100 nm particles. In the alveolar region where the airflow is very low, no deposition hotspots for NPs exist (Balásházy et al., 2008). This non-homogeneous depositoin and formation of hotspots seems to correlate with predilection sites for bronchial carcinoma (Schlesinger and Lippmann, 1978; Balásházy et al., 2003), which is further enhanced by less effective mucociliary clearance at carinal ridges.

Respiratory Tract Clearance and Disposition of NP: Nanomaterials

Once NP are deposited in the respiratory tract they will encounter physiological clearance mechanisms in the nasal–oropharyngeal, tracheobronchial, and alveolar region as described under Particle Clearance in Chap. 15. However, there are several differences that separate NP from larger particles as indicated in Table 28-2. Alveolar macrophages generally are attracted to deposited particles by chemotactic signals generated at the site of deposition, for example C5 via activation of complement by particles (Warheit et al., 1988a,b). Nanosized particles may be too small to generate such signal; studies in rodents with inhaled particles ranging from nano- to micro-size have shown that in vivo macrophage surveillance in terms of phagocytosis and clearance of nanosized particles is rather poor (Oberdörster et al., 2005; Semmler-Behneke et al., 2007; Geiser et al., 2008), so that deposited NP will interact with epithelial cells. Uptake into the pulmonary interstitium occurs, and there is evidence that some of the interstitialized NP reenter the airways, possibly the smaller conducting airways, from where they are cleared via the mucociliary escalator (Semmler-Behneke et al., 2007).

Translocation into the interstitium and subsequently into blood and lymph circulation distinguishes nanoparticles from microparticles. Fig. 28-15 depicts the blood compartment as a plenum from which any tissue/organ can be reached by circulating NP. However, the amount of NP translocating from the lung to the blood circulation and accumulation in secondary organs is very low, as results from rat inhalation studies with NP have shown, about 1% to 2% of the lung deposit (Kreyling et al., 2002, 2009; Semmler-Behnke, 2007). Long-term retention studies with radioactive iridium NPs (15–20 nm) have shown that clearance of NP in extrapulmonary organs following the initial accumulation is very efficient so that after six months, with the exception of liver and spleen, only minor amounts—below level of detection—were still present. Liver and spleen also had decreased sharply to levels of less than 0.05% of the initial lung burden. In general, the overall pattern of long-term clearance from the lungs of rats was similar to that of microparticles with retention halftimes of 70 to 90 days. Elimination of NP from the body occurred via feces (Semmler et al., 2004) and via urine for smaller NPs (see below). Despite the low translocation rates of NP from the portal of entry to secondary organs and efficient clearance, it has to be considered that a continuous exposure may result in significant accumulation in some secondary organs. Significant efforts should, therefore, be made to design studies to assess the biokinetic behavior of ENM following relevant exposures.

Biokinetic studies have shown that organs with major reticulo-endothelial system (RES) functions, such as liver and spleen, take up most circulating NP. Coating NP with PEG prevents this uptake by phagocytes and thereby the circulation time is increased, which may be desirable for medical diagnostic or therapeutic applications (Ballou et al., 2004). Other RES organs for accumulation of nanomaterials include the bone marrow and the developing organism in utero. Like the liver and spleen, the bone marrow can receive similarly high or even higher doses of blood-borne NPs (Rinderknecht et al., 2007). As a major RES organ, studies need to be designed to determine responses of stem cells in bone marrow upon interaction with ENM in vivo. A recent study in mice after administration of very high doses in drinking water of TiO₂ NP (~25 nm) for five days reported induction of DNA double strand breaks in bone marrow cells (Trouiller et al., 2009). The authors interpreted this finding as secondary to a systemic inflammatory response, multiple other organs also showed oxidative stress responses and genotoxicity. The authors did not attempt to determine the amount of TiO₂ in different organs, which would have strengthened the study.

Organs with tight endothelial junctions, in particular the central nervous system, will not likely accumulate blood-borne NPs, unless the tight blood–brain barrier is damaged or NP surface has been specially modified (Kreuter, 2004). For example, coating NP with the surfactant polysorbate 80 made it possible to deliver a drug loaded into NP across the blood–brain barrier (BBB), to the brain. Translocation across the BBB was facilitated via the endothelial low-density lipoprotein (LDL), receptor, and coating of the NP with ApoE-lipoprotein achieved the same result. The drug was either released to diffuse into the brain after NP uptake into the endothelial cells or via NP transport into the brain (Kreuter, 2004). In either case, the mechanism for drug delivery may be based on a “Trojan Horse” effect in that the coated NP mimic lipoprotein particles. This underlines again the importance of protein adsorption (corona formation) for the biodistribution of NP following uptake into the organism.

Nanomaterials and the Brain

Fig. 28-16 depicts four basic translocation pathways of inhaled NP from the upper and lower respiratory tract into the CNS. In addition to the blood circulation—which does not efficiently contribute to CNS translocation because of the very tight blood–brain barrier—there is uptake into the lymphatic system—again feeding back into the blood circulation—and transport in or next to sensory neurons from the upper respiratory tract. Perineural translocation along the olfactory nerve will deliver NP into the cerebrospinal fluid (CSF), as reported by Czerniawska (1970) and Zhang et al. (2006). To what degree transport into brain tissue may be inhibited due to a cerebrospinal brain barrier needs to be assessed in further studies. The most efficient pathway of NP translocation to the CNS appears to be via olfactory sensory neurons from the nasal olfactory mucosa directly to the olfactory bulb (Oberdörster et al., 2009). The discovery of this nose to brain connection as a NP passageway goes back to early studies demonstrating this pathway for polio virus ~30 nm transport from the nose to the brain in nonhuman primates following intranasal instillation (Brodie and Elridge, 1934; Bodian and Howe, 1941), and subsequently visualized by transmission electron microscopy (TEM), with silver-coated gold NP (50 nm) instilled intranasally into squirrel monkeys (De Lorenzo, 1970). More recent inhalation studies with insoluble carbon NP (~36 nm) and poorly soluble Mn-oxide NP (~30 nm) confirmed the olfactory nerve system as an efficient mechanism for airborne NP to reach the CNS (Oberdörster et al., 2004; Elder et al., 2006). Inflammatory responses in the olfactory bulb and several brain regions were induced by 12 days of manganese oxide exposure at concentrations that are lower than those experienced by arc welders after exposure to a mixture of nanosized particles—including manganese oxides. Other nasal instillation studies in mice, applying enormous high doses, reported oxidative stress and inflammatory responses (Wang et al., 2008a,b). Results of epidemiological studies of impaired cognitive function and of neurodegenerative brain pathology associated with exposure to traffic-related particles raised the question as to whether ambient UFPs as constituents of urban air pollution may be etiologically involved (Calderon-Garciduenas et al., 2011; Weuve et al., 2012). Whether such etiological link exists is still to be investigated. The plausibility of a respective hypothesis is supported by findings that loss of olfactory function has been described as a common characteristic of neurodegenerative diseases, that is, Parkinson, Alzheimer, and Huntington disease (Barrios et al., 2007; Doty, 2008; Kovács, 2004; Moberg and Doty, 1997).

In the same study by Trouiller et al. (2009) mentioned before, the authors had also exposed pregnant mice to TiO₂ NP in drinking water for 10 days. The offspring had a higher frequency of DNA deletions in the retina when examined at 20 days of age. Due to the extraordinarily high dose of 500 mg/kg and the lack of verifying TiO₂ in fetal tissues, these results have to be interpreted with caution. Other studies evaluating exposure of the fetus to NP size-dependent translocation across the placenta in mice, with smaller quantum dots transferring more readily than larger ones (Chu et al., 2010). Takeda et al. (2009) injected TiO₂ NP subcutaneously in pregnant mice (100 μg/mouse on each of four days). They identified TiO₂ NP in testis and brain of the six-week-old male offspring. The authors also observed reduced daily sperm production and apoptotic cells in the olfactory bulb of these offspring. Results of this bolus-type injection study maybe viewed as a proof-of-principle study for demonstrating placenta penetration because of the unphysiological mode of delivery of a high dose and high dose rate; however, the same group had reported in a mouse inhalation study with diesel exhaust the presence of UFPs in the brain of offspring and increased apoptosis of Purkinje cells in the cerebellum (Sugamata et al., 2006).

Elimination of Nanomaterials

Elimination pathways for ENM from the body (Fig. 28-15) include mainly feces and urine. Urinary excretion is restricted to nanostructures <5.5 nm in size for metal-based NP (Choi et al., 2007), limited by the hydrodynamic diameter of the particle that may change in circulation due to protein adsorption. This general principle appears to be different for circulating fibrous structures of ENM. Singh et al. (2006) and Lacerda et al. (2008) reported that even 20 to 30 nm diameter MWCNT with length between 0.5 and 2 μm were collected in urine of rats following intravenous application. They explain this phenomenon by a hydrodynamic lining up of nanotubes so they will pass through glomerular pores. The fecal excretory clearance pathways consist of several inputs: One is mucociliary clearance of deposited particles from the airways into the GI tract; another is via hepatobiliary clearance of blood-borne ENM via liver and bile into the small intestine. This elimination pathway is suggested by results from Semmler-Behnke et al. (2007), and it is also a well-known excretory path for heavy metals in the blood (Clarkson et al., 1988).

With regard to the CNS, no data on ENM elimination are available yet. It is conceivable that the CSF via its connections to the nasal lymphatic system (Czerniawska, 1970) and to the blood circulation could be an excretory pathway for the brain, which needs to be investigated in future studies. Indeed, Segal (2000) concluded from their review on CSF barriers that the CSF may both act as a compartment for distribution of substances to many brain regions, but also can act as an elimination route for waste products into the blood circulation because the brain has no lymphatics.

Another clearance pathway of deposited ENM in the lung involves translocation via interstitium or lymph to the pleura and subsequent elimination via lymphatic openings (stomata) on the parietal pleura to mediastinal lymph nodes from where NP may enter the blood circulation via the right and left thoracic ducts. This pathway is of particular importance for fiber-shaped ENM because the size of the parietal stomata prevents efficient clearance of structures >10 μm in length. Indeed, collection of lymph from the right thoracic duct of dogs that had been dosed intrabronchially with amosite asbestos showed only fibers <9 μm in length appearing in the collected lymph (Oberdörster et al., 1988). As a consequence, the interaction of the retained longer fibers in the pleural cavity with mesothelial cells induce inflammatory and granulomatous responses and in long-term potentially mesothelioma (Donaldson et al., 2010). The following section discusses experimental in vivo data from bolus-type and inhalation studies of the toxicology of CNT that confirm the possibility of a mesotheliogenic response, but also caution to extrapolate directly to realistic inhalation exposures without further verification. A critical appraisal of the available in vivo study concludes this section.

Bolus-type Exposures

As described in the previous paragraphs, bolus-type delivery has been the preferred method of dosing to identify as a first step of the potential hazard of ENM. The advantages and disadvantages need to be considered when interpreting the results, in particular the rationale for the administered dose levels causing potential bolus associated responses should be critically considered. Several doses covering a reasonable range, also including a positive or negative control material for ranking purposes, should be part of a well-designed study. Knowledge about actual—if available—or anticipated exposure levels should be used to justify a selected dose range. Key is a careful physicochemical characterization of CNTs with respect to their original specification (pristine state) as well as any modification due to treatment (dispersants, sonification) prior to administration. Specifically, the presence of impurities (eg, metals, organics, amorphous carbon) needs to be considered.

Early intratracheal instillation studies in rats and mice with SWCNTs were performed with very high doses (up to 1.25 mg/rat and 0.5 mg/mouse), causing severe lung injury with rapid persistent granuloma formation including mortality (Warheit et al., 2004; Lam et al., 2004). Doses were so high to induce even physical blockage of conducting airways. Pharyngeal aspiration of lower doses with purified SWCNTs in mice at lower doses caused interstitial fibrosis and granulomatous lesions, possibly due to the existence of large agglomerates in the administered suspension (Shvedova et al., 2005). Subsequent aspiration studies in mice confirmed these findings in mice, showing also that the responses were not affected by the metal content; however, the degree of dispersion influenced responses such that administration of the same mass dose of well-dispersed SWCNTs caused greater and more persistent responses than poorly dispersed SWCNTs (Shvedova et al., 2008; Mercer et al., 2008).

MWCNTs delivered to rats by intratracheal instillation (0.5–5 mg/rat) also induced inflammatory, granulomatous, and fibrotic responses; these responses were lower when the material was ground such that the length of the MWCNTs was reduced from 6 to 0.7 μm (Muller et al., 2005). Another study compared pulmonary responses to purified MWCNTs in mice when given either by instillation (50 μg/mouse) or inhalation (32.6 mg/m³, 6 h/day, up to 15 days, estimated to result in 70–210 μg in the lung) (Li et al., 2007). Despite the higher lung dose following inhalation, only moderate effects were induced, whereas instillation resulted in severe inflammation and alveolar wall destruction, demonstrating a stark difference between responses induced by the two modes of dosing with the physiological inhalation exposure avoiding the high bolus-related response. As already mentioned before, the results of both the Li et al. and Shvedova et al. studies have to be interpreted with caution because of the uncertainty of the deposited doses in the lung following inhalation.

Other studies with instilled high doses (up to 1.75 mg/rat) of MWCNTs in rats reported transient inflammatory responses with persistent alveolar wall thickening (Liu et al., 2008). Transient inflammation and granulomatous responses in rats were also reported by Kobayashi et al. (2010) at lower instilled MWCNT doses, whereas Porter et al. (2010) observed a rapid and persistent fibrotic response in mice following aspiration of purified MWCNTs in mice (10–80 μg/mouse) which was confirmed by morphometric (lung tissue) analysis by Mercer et al. (2011). Additional evaluations of MWCNT responses in rats and mice confirmed inflammatory, granulomatous, and fibrotic responses following instillation or aspiration dosing (Porter et al., 2010; Aiso et al., 2011; Han et al., 2010).

Translocation of aspirated MWCNT to the pleura following aspiration in mice was also reported (Porter et al., 2010) including penetration into the intrapleural space of mice following aspiration (Mercer et al., 2010). A six-hour inhalation exposure of mice to a very high concentration of 30 mg/m³ confirmed that inhaled MWCNT can reach the subpleural wall (space between peripheral alveoli and visceral pleura) are localized within subpleural macrophages (Ryman-Rasmussen et al., 2009a). In summary, bolus-type delivery of CNTs to the respiratory tract of rats and mice revealed induction of dose-dependent significant inflammatory, granulomatous, and fibrogenic responses; they showed also that MWCNTs can reach subpleural and intrapleural sites.

Other bolus-type studies delivered MWCNTs into the peritoneal cavity (intraperitoneal injection, or i.p.)—as a surrogate for the mesothelial lining of the pleural cavity—in order to expose mesothelial cells directly for assessing a mesotheliogenic response. Takagi et al. (2008) found, indeed, mesothelioma induction one year following very high i.p. doses in p53 heterozygous mice, which was similar to the positive crocidolite asbestos control. A follow-up study with lower i.p. injection of 50 μg/mouse confirmed this finding (Kanno et al., 2010). Similarly, Sakamoto et al. (2009) induced peritoneal mesothelioma in rats, following high-dose MWCNT injection intrascrotally. Poland et al. (2008) also suggested an asbestos-like response after inducing mesothelial granuloma with amosite asbestos and with long straight but not with short MWCNT two weeks after i.p. injection in mice. Follow-up i.p. studies showed the importance of lymphatic (Fig. 28-17) clearance of fiber structures from the pleural cavity (Murphy et al., 2011): Lymphatic stomata (Wang et al., 1975) in the parietal pleura with diameters of 3 to 10 μm in rodents (Shinohara et al., 1997) act as an effective clearance pathway for shorter fibers, whereas longer ones will be retained and induce inflammatory, oxidative stress responses progressing to mesothelioma. Such length-dependent plural retention and lymphatic clearance was demonstrated by Murphy et al. (2011) after using i.p. injection of different types of MWCNTs and other nanotubes. Only long but not short or entangled MWCNTs caused significant mesothelial cell proliferation and granuloma formation and fibrosis. These i.p. injection studies in rodents clearly show the potential of CNTs, specifically MWCNTs, to induce severe adverse length-dependent effects at mesothelial sites once they reach the pleural cavity. Appropriate inhalation studies in rodents need to be designed to determine the translocation of inhaled, generally highly agglomerated airborne CNTs from deposition sites in the lung to the pleura and to confirm the findings from bolus aspiration studies with well-dispersed CNTs.

Inhalation Studies

Relatively few inhalation studies with CNTs in rodents have been reported (Table 28-8). Most of these were of short duration (hours to a few days), and only two were designed as subchronic 13-week inhalation studies (Ma-Hock et al., 2009; Pauluhn, 2010) and two as four-week inhalation studies (Oyabu et al., 2011; Morimoto et al., 2012a). The value of short-term inhalation studies for risk characterization can be questioned because such exposures to CNTs at realistic low concentrations are not likely to reveal certain chronic effects if long latency periods are involved. Although unreasonable, very high concentrations with short exposures maybe helpful for purposes of hazard identification and toxicity ranking—which could be achieved at less cost and effort by bolus-type studies—they are not useful for deriving a no observed adverse effect level (NOAEL) or for quantitative risk assessment. Thus, most meaningful and best justified for the risk assessment process would be a subchronic multiconcentration (minimum three concentrations) study with sufficient postexposure observation. However, short-term, even one-term, exposure to a relevant concentration is very useful for dosimetric purposes when determining the biodistribution from deposition sites in the lung to secondary organs. Of course, a prerequisite is the availability of sensitive detection methods, such as radioactive labeling (Kreyling et al., 2009) or use of a nonleachable catalyst firmly attached to or imbedded within the CNTs.

The choice of the mode of inhalation exposure—whole-body or nose-only—is up to the investigator. Generally, for long-term chronic exposures, whole-body is preferred because the tight confinement of rodents during nose-only exposures induces additional stress, despite careful adaptation prior to starting exposures (Rothenberg et al., 2000; Fawcett et al., 1994; Pare and Glavin, 1986; Udelsman et al., 1993; Sistonen et al., 1992). The disadvantage of whole-body exposure is the contamination of the animals’ fur with aerosolized CNTs, which may lead to additional oral intake due to preening. However, the additional dose is very low, and GI-tract exposure is part of any inhalation exposure due to bronchial clearance of the deposited material.

Most of the CNT inhalation studies were performed with MWCNTs; only two used SWCNTs. Exposure concentrations ranged from 0.1 to 32.6 mg/m³ either one concentration only up to four concentrations in a study, and duration from six hours to 13 weeks. Outcomes (Table 28-8) range from no significant effects to severe pulmonary inflammation/oxidative stress responses; one study reported systemic effects including cytokine and oxidative stress responses in the spleen (Mitchell et al., 2007). Given the diversity of the used CNTs and the different exposure methods, aerosol preparation, the airborne CNT characteristics, and different endpoints, the wide range of responses should come as no surprise. Although these studies show the potential of inhaled SWCNT and inhaled MWCNT to cause lung injury, transiently or more persistently, the two subchronic multiconcentration 13-week studies allow a more complete assessment of the toxicity of MWCNTs that could be used for deriving occupational exposure limits (Pauluhn, 2010).

Although bolus-type delivery did result in pleural effects and translocation of MWCNT toward pleural sites as discussed before, no pleural effects were seen in the 13-week inhalation studies, except pleural thickening at the two highest exposure concentrations (Pauluhn, 2010). Movement of inhaled MWCNT to subpleural sites was observed by Ryman-Rasmussen et al. (2009a) when individual fibers and very fine agglomerates (160–210 nm) were generated from a liquid suspension following a single six-hour very high concentration exposure of 30 mg/m³. However, pleural fibrosis or granuloma did not develop up to 22 weeks postexposure. The small size of the aerosol as well as an extremely high dose may have been a factor in a deeper peripheral lung deposition. A follow-up six-hour study by Ryman-Rasmussen et al., (2009b) with inhaled MWCNT in an ovalbumin (OVA), allergic mouse model found a synergistic effect; MWCNT alone did not induce fibrosis in nonsensitized mice. This result is in contrast to their earlier study where a six-hour exposure to a three-fold lower concentration did induce subpleural fibrosis in mice (Ryman-Rasmussen et al., 2009a). The larger aerosol size in the second study may have prevented significant deposition in the peripheral lung.

Regarding a comparison between MWCNT- and SWCNT-induced pulmonary or secondary organ effects, the only medium-term (four weeks) SWCNT inhalation study in rats (Morimoto et al., 2012a) did not observe any effect; however, estimated lung burdens were just 1.8 and 8.8 μg, about the same as the lowest concentration in the Pauluhn (2010) study with MWCNTs that also did not show any effect (NOAEL).

Critical Appraisal of CNT In Vivo Studies

Several intrapleural injection studies with MWCNTs in mice have indicated that CNTs have a mesotheliogenic potential, discussed in the previous section. In addition, intratracheal instillation studies in rats and oropharyngeal aspiration studies with MWCNTs and SWCNTs have confirmed a pulmonary inflammatory and oxidative stress inducing potential. These results have raised concern that CNTs may represent and give rise to another case of asbestos toxicity and pathology, in particular insidious because of the decades-long latency period from exposure to manifestation of mesothelioma. Controversy exists regarding the applicability of results of intracavitary injection studies for regulatory purposes. European regulations for synthetic mineral fibers exonerate those from being classified as a carcinogen if it can be shown that the material fulfills one of the following conditions (EC Directive 97/69/EC, Nota Q, 1997):

• a short-term biopersistence test by inhalation has shown that the fibers longer than 20 μm have a weighted half-time less than 20 days, or

• a short-term biopersistence test by intratracheal instillation has shown that the fibers longer than 20 μm have a weighted half-time less than 40 days, or

• an appropriate intraperitoneal test has shown no evidence of excess carcinogenicity, or

• absence of relevant pathogenicity or neoplastic changes in a suitable long-term inhalation test.

This regulation is based on a number of experimental studies going back to Stanton et al. (1981) and Pott et al. (1987) that showed induction of tumors when different types of asbestos and persistent synthetic mineral fibers were directly injected into the pleural or peritoneal cavity. However, use of this test for regulatory purposes has not been established in the United States; the test has been criticized because of the massive doses being used leading to fibrotic reactions thereby interfering with and overwhelming physiological defense mechanisms. As to whether present European regulations will be applied to CNTs is questionable because they have been issued for man-made vitreous (silicate) fibers and the high bolus doses and associated caveats should be considered.

Still, results of CNT intracavitary injection studies summarized in this section have defined the potential of CNTs to induce mesothelioma, depending on their dimension (>10 μm) and their agglomeration state. Longer-term inhalation studies need to confirm these findings; presently, two subchronic inhalation studies with agglomerated MWCNTs did not confirm a mesotheliogenic response, although high concentrations in the mg/m³ range showed some pleural thickening. Additionally, acute short-term inhalation studies using relevant exposure concentrations of long (>10 μm) and short (<5 μm) MWCNTs are also needed in order to determine the biodistribution of physiologically administered CNTs (inhalation) to secondary organs. Biodistribution based on bolus-type delivery using pretreatment with dispersants and lengthy sonication may be different quantitatively and qualitatively.

Longer-term inhalation studies with intentionally well-dispersed CNTs are not available, although two four-week studies using aerosolization from an aqueous suspension of short MWCNTs and short SWCNTs in rats with use of dispersants and two acute six-hour studies with short MWCNTs in mice using dispersant suspension resulting in smaller aerodynamic diameters reported more peripheral, even subpleural, deposition, but no pleural effects. This appears to be in agreement with results from intraperitoneal injection studies, which did not show an adverse effect of short, straight MWCNTs because of effective removal by lymphatic clearance.

Obviously, given the importance of the physicochemical properties of CNTs for inducing adverse effects, it is of utmost importance to determine these properties, in particular as they appear in the airborne state at sites of human exposures, at occupational sites, or for the consumer. Adding dispersants for testing purposes will change surface properties; conceptually, inhalation studies in experimental animals for purposes of hazard identification should mimic human exposure conditions with regard to airborne size distribution. Of course, differences in respirability between humans and rats/mice have to be considered, and—if necessary—adjustments be made without use of surface altering dispersants or harsh physical treatment.

Appropriately designed multiconcentration subchronic inhalation studies, including a longer recovery period, are essential for deriving NOAELs; results can be used as basis for deriving occupational exposure levels (OELs) by applying rodent/human dosimetric adjustments (Fig. 28-18). Using results from bolus-type studies is difficult and raises questions, although national institute for occupational safety and health (NIOSH) (2011) has combined results of fibrotic responses from diverse bolus-type and inhalation studies to derive a provisional recommended exposure level (REL) of 7 μg/m³. This REL is based on dose–response data from the available studies with bolus-type and short-term inhalation exposures and a subchronic inhalation study. Differences in lengths, use of dispersants, impact of hugely different dose rates between the studies were not considered. NIOSH (2011) suggested, though, that the mass-based REL may not be sufficiently sensitive for detecting CNTs by air sampling and calls for research to determine the most sensitive dosemetrics.

Collectively, the available in vivo toxicological database of CNTs have identified a fibrogenic and mesotheliogenic effect (hazard identification). No pulmonary tumorigenic effect has been reported so far, in contrast to studies with biopersistent asbestos. An asbestos-like mesothelioma response has only been observed following direct i.p. injection of MWCNTs; confirmation following inhalation is still lacking although it is conceivable that the old paradigm of dose–dimension–durability for fiber carcinogenicity applies also to CNTs: they are quite durable (though some studies have now reported degradation of carboxylated SWCNT by peroxidases in vitro, see below), and their dimensions in terms of length and agglomeration/entanglement appear to be important for pleural effects. Of course, other properties (chemistry, surface reactivity) may play a role as well. Thus, as long as no conclusive data regarding carcinogenic effects of realistic exposure to CNTs are available, exposure should be avoided with appropriate measures (ventilation, filtration, personal protective equipment). There is an obvious and urgent need to perform additional long-term (subchronic to chronic) inhalation studies to assess the carcinogenic potential of CNTs and derive protective OELs.

Biological Degradation of Carbon Nanomaterials

CNTs have been generally regarded as stable nondegradable materials, which has important implications for long-term health effects following inhalation into the lungs (Donaldson et al., 2006). Recently, however, SWCNT degradation has been observed in acellular assays that simulate the phagolysosome of macrophages, but only if the tubes have been surface carboxylated, which introduces collateral defects in the side walls (Liu et al., 2010c). Carboxylated tubes are also susceptible to biodegradation by exposure to hydrogen peroxide and horseradish peroxidase (Allen et al., 2008, 2009; Kagan et al., 2010) or hypochlorite and the mammalian enzyme myeloperoxidase (Kagan et al., 2010). Kotchey et al. (2011) also reported that graphene oxide, but not reduced graphene oxide, is susceptible to oxidative attack by hydrogen peroxide and horseradish peroxidase. These observations may enable design of safer carbon materials (Yan et al., 2011) that are potentially biodegradable in order to minimize adverse environmental and human health impacts. However, degradation of CNTs in vivo is still to be confirmed.

Although numerous in vitro and in vivo studies have reported toxic effects of different ENMs, results are often conflicting and hazards are overstated due to poor understanding and knowledge about the appropriateness of applied doses and concentrations and their relevance to realistic exposures. This issue will be further discussed later. A conceptual testing paradigm is shown in Fig. 28-19. In order to perform risk assessment or to derive OELs, exposure and hazard data are required. To identify and characterize a hazard, in vitro and in vivo studies will be useful, and results should be derived via well-designed dose–response relationships. Key considerations for designing such studies include careful physicochemcial characterization of the ENM to be tested, justification of the method(s) of dosing, selection of target cells, tissues or animal species, and appropriate endpoints. Reference materials may also be included in order to rank unknown ENMs against a known positive and/or negative control. Whereas in vitro data alone permit toxicity ranking, appropriately performed in vivo studies (using inhalation when dealing with airborne materials) will allow a full risk assessment because exposure and dosimetric extrapolation to humans can be performed. In the long-term, in silico studies may be developed to assess a hazard, and in even a more distant future to predict human risk. Ideally, knowledge about anticipated human exposure would be necessary to inform both animal exposure studies as well as in vitro dosing.

In vivo animal studies obviously are not generally predicting human exposures: the lower loops in Fig. 28-19 are referring to dosimetric correlations between in vitro/in vivo and in vivo animals/in vivo humans, going in both directions with the goal to (a) inform the design of in vivo animal studies using available human exposure data, (b) use exposure–dose–response information from animal studies to compare with human data. The upper one directional animal to human loop is referring to extrapolating effects and mechanisms from relevant (based on dosimetric relevancy) animal studies to humans, with the goal to be predictive for deriving “safe” exposure levels or OELs; thus, an animal study can be used to predict human equivalent exposure concentrations or OELs. An example for this approach is the recent subchronic rat inhalation study with MWCNT by Pauluhn (2010).

The following paragraphs focus on concepts and results of in vivo experiments as basis for use in the risk assessment process using as example CNTs. However, a need for in vitro studies should be stressed as far as uncovering underlying mechanisms of effects are concerned, provided that the caveats pointed out above are considered. In addition, despite these limitations, in vitro studies are useful for toxicity ranking of nanomaterials for the purpose of hazard identification (Rushton et al., 2010). In contrast, the design of in vivo studies allows the full evaluation of exposure–dose–response relationships, which is necessary for the process of risk assessment (Fig. 28-19).

Some important points should be considered when designing exposure–dose–response studies with CNTs in experimental animals. The major exposure route for CNTs to be expected is via inhalation at manufacturing sites and during distribution and usage (handling, refilling, disposal), although ingestion and dermal exposures also occur but have not received as much attention (Fig. 28-15). For medicinal applications, injection is an important route of exposure requiring specific awareness with respect to assuring desired beneficial (pharmacological) outcomes yet avoiding undesirable (toxicological) responses (Kolosnaj et al., 2010). For example, Yang et al. (2010) concluded from their studies with SWCNT that the desired pharmacological target organelle for drug delivery by SWCNT are the cell lysosomes, whereas the mitochondria are the target organelles for SWCNT toxicity. These aspects will not be discussed further.

In Vitro Dosimetry

The importance of dose rate for bolus-type delivery of ENM in vivo vis-à-vis inhalation has been stressed in previous sections. As most ENM toxicity studies are performed using in vitro assays, which are generally short-term, dosing-related questions are highly relevant. Teeguarden et al. (2007) pointed out that the dose received by the cell is a function of colloidal dynamics or “particokinetics” in the culture medium governed by diffusion and settling phenomena, which in turn is governed by particle and media properties that include particle size, agglomeration/aggregation state, shape, density, and charge. This is very different from chemicals dissolved in culture medium and requires a thorough evaluation of particle dosimetry in vivo. At equal mass concentrations in the medium, the magnitude of the cellular dose of ENM will differ significantly from those implied by the media concentration. This can give rise to significant misinterpretation of responses and is likely to contribute to differences in results between different laboratories. For example, particles of the same material but of different sizes will settle onto cells in culture at different rates; and particles of the same size but different densities will also settle differently so that the time for particles to reach the cell layer at the bottom of a dish can vary widely. Similar to the aerodynamic behavior of airborne particles, the hydrodynamic behavior in culture medium expressed as their transport velocity shows a minimum of movement for intermediate sizes where the combination of gravitational settling and diffusional movement is lowest.

Fig. 28-20 illustrates the different mechanisms that determine the in vitro particle kinetics for particle sizes from 1 nm to 1000 nm (Hinderliter et al., 2010). A typical U-shaped form depicting faster transport at both ends of the range of sizes and a slowest transport for particles around 20 to 60 nm is the result of the competing forces and particle and media characteristics. Temperature, mixing, and advection also add to the complexity of particle movement in the media. Teeguarden’s group developed an In Vitro Sedimentation Diffusion and Dosimetry (ISDD) model for predicting the in vitro behavior and cell doses of particles. Fig. 28-21 shows the results of example calculations from this model. It indicates that for large TiO₂ particles (500 nm) the dose received by the cells in a dish with 3.1 mm media height is very fast. In contrast, 30 and 50 nm particles have a long deposition rate; only 50% of the media dose has reached the cells by 24 hours. TiO₂ NP sizes in between still have not reached the full dose by 24 hours, whereas 200 nm TiO₂ NP are fully settled by 10 hours.

Results of the ISDD prediction shown in Fig. 28-21 have to be verified by experiments. The value of this model lies in the clear separation between exposure (concentration in the cell medium), the deposited dose on the cell surface, and—depending on cellular uptake—the cellular dose. These doses can be expressed by different metrics: mass, number, or surface area of the nanomaterial. Knowledge about the time to deposit a certain dose allows to consider dose rate as a determinant for responses, as was discussed for in vivo exposures (Fig. 28-8). In vitro dosing is often perceived as a bolus-type delivery, similar to in vivo instillation or aspiration studies. However, as the results in Fig. 28-21 show, in vitro dosing with NP is not necessarily equivalent to an instantaneous bolus. The ISDD model allows to express the deposited dose per unit cell surface area of cultured epithelial cells of the respiratory tract and thereby (i) compare responses to the same in vivo dose per unit cell surface (based on MPPD model, Fig. 28-13), (ii) use in vivo-derived doses from a given exposure concentration to design in vitro studies that include realistic doses, and (iii) use in vitro dosimetry results and response to verify in vivo. Likewise, doses applied in vitro to cells from secondary organs (eg, CNS, bone marrow) can be estimated by the ISDD model and better justified by results from in vivo biokinetic studies with respect to doses distributed to these secondary organs. An important difference, however, between in vivo inhalation and in vitro cell exposure is the pristine state of inhaled nanomaterials, whereas for in vitro exposures nanomaterials are either treated with dispersants to separate the particles and achieve spatially uniform dosing, or else culture media constituents (serum proteins) will adsorb to the NP surface to aid in their dispersion. Surface modification is known to affect responses of ENM (Haniu et al., 2011), and this possibility needs to be considered when designing and interpreting in vitro results.

An alternative in vitro methodology that avoids pretreatment of NP with surfactants or proteins prior to testing is exposure via an air liquid interface (ALI) system (Savi et al., 2008; Holder et al., 2008; Lenz et al., 2009; de Bruijne et al., 2009; Volckens et al., 2009; Tang et al., 2011; Xie et al., 2012). Respiratory tract epithelial cells are grown on porous culture cell inserts and exposed to NP generated the same way as done for in vivo inhalation studies. Before exposure, culture medium is withdrawn so that only the basal cell surface is in contact with the medium and deposition of the aerosolized NP on the apical cell surface simulates realistic in vivo inhalation exposure. Comparative results between conventional in vitro and ALI exposure seems to confirm the apparent advantage and greater sensitivity of the ALI system (Holder et al., 2008; Volckens et al., 2009). However, transition of cell cultures from growing in suspension to the ALI system for exposure is likely to cause changes in gene expression, which needs to be considered using appropriate control groups (Ross et al., 2007).

Predictive Toxicology

The NNI developed a comprehensive Environmental, Health, and Safety Research Strategy (NanoEHS research, NNI, 2011) in which urgent research needs for nanomaterial measurements, human exposure assessment, human health, environmental health, risk assessment and risk management methods, and for the input of informatics and modeling for NanoEHS research are discussed. This initiative of the US Government’s interagency program for coordinating research and development and for enhancing communication and collaborative activities in nanoscale science, engineering, and technology emphasizes the NAS risk assessment paradigm (1983) as the basis for NanoEHS research. As illustrated in Fig. 28-19, data about hazard and exposure are key for the risk assessment process. Critical elements for hazard identification based on toxicity testing of ENM are detailed information about their physicochemical properties prior to any experiments, the selection of appropriate target cells, validation of in vitro assays in terms of correlation and relevancy to in vivo results, the inclusion of biokinetics in the design of in vivo studies, and the inclusion of realistic doses in the design of dose–response in vitro and in vivo studies. Biokinetic information is crucial to identify potential secondary target organs based on significant accumulation of ENM. In order to determine relevant doses, information about exposures occurring at anticipated real-world exposure scenarios, either measured or estimated, is essential. With this information, exposure–dose–response (in vivo, rodents) and dose–response (in vitro, target cells) relationships can be established to both characterize a hazard and assess a risk (see below). In terms of toxicity ranking, it is desirable to include reference materials in the study design so that a hazard can be expressed relative to a positive (high hazard) or negative (low hazard) control ENM. Mechanistic information discovered through in vitro assays will further aid in the characterization of hazard, provided that the mechanism is operative at relevant doses/concentrations. High-dose and high dose rate-induced mechanisms may not be considered relevant and may give rise to improper classification. An awareness of dosimetry-related aspects is of highest importance (Fig. 28-19) for the risk assessment process, which often gets lost or is ignored because of a misconception of risk being analogous to hazard. Risk is a function of hazard and exposure, and neither aspect alone can determine risk.

The immediate goal of ENM toxicity testing is hazard characterization and the establishment of respective predictive tests. These should ideally be based on validated in vitro studies, for example high throughput in vitro assays, so that ethically controversial animal studies can be replaced if they only serve the purpose of hazard identification. A distant goal is the use eventually of in silico models. However, at present, animal studies are still indispensable for obtaining crucial information about long-term effects of ENM and to obtain results for risk assessment. For example, available data from subchronic (three months) inhalation studies (Ma-Hock et al., 2009; Pauluhn, 2010) provide crucial information for deriving a NOAEL or lowest observed adverse effect level (LOAEL) that can be used for dosimetric extrapolation of OELs for humans.

Fig. 28-18 depicts a conceptual approach for deriving a human equivalent concentration (HEC) from results of a long-term rat inhalation study. The HEC is based solely on the equivalency of the retained dose, expressed per unit mass, or epithelial surface area of the lung. The retained dose can be expressed by different metrics, for example, rather than using the mass of ENM, as is conventional, the surface area or even number of particles might be used. Pauluhn (2010) used the metric of particle volume retained in alveolar macrophages to estimate an OEL from the experimentally found NOAEL in a three-month rat inhalation study with MWCNT. As was done in that study, it is highly desirable that the design of a rat study includes the measurement of the retained lung burden; if this is not available, the retained dose can be estimated using a particle deposition and retention model with inputs of rat-specific breathing and airway parameters and the aerosol characteristics of the ENM used in the rat study. The same aerosol characteristics and human-specific airway and breathing parameters then as inputs to a human model are then used to determine the HEC. It should be noted that this dosimetric extrapolation should not necessarily imply that the response in humans is the same as in rats. Differences in species sensitivity may require the use of an additional safety factor. The advantage of using the retained dose as the basis is that other particle size distributions specific to human exposures at a workplace can be used instead of the rat aerosol. (Oller and Oberdörster, 2010).

Transition, Human—Eco-nanotoxicology

The goals of nanotoxicology are to identify and characterize a hazard of ENMs for purposes of risk assessment for humans and the environment require a highly multidisciplinary team approach, covering expertise in toxicology, biology, chemistry, physics, material science, geology, exposure assessment, PBPK and fate and transport modeling, and medicine is necessary to develop testing strategies, establish toxicity ranking, determine “safe” exposure levels and derive preventive exposure guidelines. As depicted in Fig. 28-2, exposures throughout the lifecycle of ENMs, from their source to their disposal, have to be considered. Dispersion and intermediate transformations in air, water, food, and soil are important modifiers of ENM-receptor interactions. Identifying underlying mechanisms of such interactions will aid in assessing risk and establishing preventive measures. Developing predictive models is a distant goal for both human and eco-nanotoxicology.

Environmental Uses and Exposures to Nanomaterials

As with many chemicals in the marketplace, it is estimated that a portion of the nanomaterials used in industry and consumer products will enter the greater environment during some part of their life cycle, either through waste during production or through product use. The exponential increase in use of ENMs in a multitude of industries and consumer products has been documented by the Woodrow Wilson Center’s project on Emerging Technologies, which as of March 2011 estimated that self-identified nanomaterial products had grown from 54 to over 1300 in five years. This list likely underestimates the number of uses of nanomaterials as it only includes products easily identified as containing nanomaterials and does not include other known uses of nanomaterials including industrial catalysts, wastewater treatment, in addition to environmental cleanup technologies and pesticides, which would directly place nanomaterials into the environment. Most of the uses identified are in cosmetics, clothing (where nanomaterials are often added for antimicrobial and antiodor functions), personal care products and sporting goods. The most common nanomaterials included in these products are silver and carbon nanomaterials. Other chemicals used in personal care products, cosmetics, and clothing have been found to wash into the wastewater treatment system and end up in the aquatic environment either directly from the treatment facilities or through land applications of biosolids from wastewater treatment plants. It is anticipated that some of the nanomaterial components may have the same fate, and indeed a few documented cases indicate this may be possible (Wang et al., 2012) but a majority of the estimates of potential environmental release of these chemicals is from modeling based on data from other compounds without regard to specific properties of nanomaterials compared to their bulk counterparts (Mueller and Nowack, 2008, Gottschalk et al., 2009, 2010). These studies estimate that concentrations in the environment can range from ng/L to ug/L with many agglomerating into larger than nanosized units and much of the NP mass binding to solids and removed into biosolid waste. However, despite these removal potentials, it has been shown that NPs can be emitted through the wastewater process in the NP size range in significant quantities (Westerhoff et al., 2011).

Nanomaterials directly applied to a particular ecosystem, such as those for cleanup of environmental toxicants or as part of a pesticide formulation may also lead to exposures. Nanoscale zero-valent iron and titanium dioxide are currently being evaluated as a removal technology to treat groundwater systems and some surface water or soil systems for metals, arsenic, organic chemicals such as chlorinated solvents, polychlorinated biphenyl’s (PCB’s), and benzenes (Elliot and Zhang, 2001; Chen and von Mikecz, 2005; Quinn et al., 2005). Poly(ethylene) glycol-modified urethane has been proposed for the sequestration and enhanced bioavailability for degradation of organic contaminants (Tungittiplakorn, 2005). Pesticide formulations containing carbon-based nanomaterials and others have been sent to the US environmental protection agency (EPA), for evaluation and are assumed to be in development.

Ecological Risk Assessment of Manufactured Nanomaterials

One of the unusual aspects of the field of nanotechnology is the way in which the science of the risk of these materials to the environment has been instigated relatively early in the adoption of the technology. For most other contaminants, the risk assessment has largely been retrospective often decades after the chemical has been in the marketplace and is now distributed across the globe into many environmental compartments. The goal of this science to date has been to determine how a particulate form of a chemical in a nanosize range influences its toxicity versus its larger bulk counterparts. A further goal is to determine how small changes in surface chemistries, size, shape, and structure affect the interaction with organisms in the environment with the goal of informing industry on safe development. However, there are issues in assessing the risk of nanomaterials that are unique to this class of potential contaminant, and therefore the process of assessing risk also has unique considerations (some that have been mentioned in dealing with human health considerations). The assessment of ecological risk is also complicated that human health assessments as there are more species involved and higher level impacts on populations, communities of organisms and ecosystem function need to be considered. The methods to evaluate the potential impacts of nanomaterials at higher levels of organization (community, ecosystem) have not been developed as is the case with many other chemicals.

The basic framework used for ecological risk assessment (Fig. 28-23) (EPA EPA/630/R-95/002F) involves problem formulation, analysis, and risk characterization. Problem formulation involves selecting endpoints of concern and developing a plan for determining impacts on important endpoints (such as a species decline, ecosystem changes). Analysis involves evaluating the potential for exposure to stressors and potential effects at those exposure levels. Risk characterization involves taking the information on important endpoints and the analysis of exposure and effects data to determine the importance of a stressor. For nanomaterials most of the ecological risk assessment research has been conducted as the analysis of effects of a limited number of commercially available materials, using traditional acute single-organism mortality endpoints of a few select species and with little information regarding sublethal types of effects or other endpoints of concern at the community or ecosystem level. In addition, the concentrations of exposures are much higher than what is considered to be a probable environmental level. As such it is difficult to draw conclusions of the actual risk of many of even the most common materials and the conclusion is often that these particles may not cause any impact as no acute toxicity is seen. Nanomaterials may also be transformed within the environment, and therefore the toxicity of the initial nanomaterial may not provide a complete idea of the toxicity over the lifetime of the material (Nowack et al., 2012). Part of this is due to the relative newness of this field, the particulate nature of the potential toxicants, and other challenges surrounding the methods for determining toxicity, methods for measuring potential exposures, and uncertainty in estimates of how the materials will be used in the future marketplace. Agencies involved in protecting public and environmental health in individual countries as well as international agencies are grappling with methods to develop a structure to assess risk for all nanomaterials and make predictions regarding new ones as they come on the marketplace. This assessment may take time as methods development needs to progress with the field of nanotechnology as new materials are being developed. The current state of the field is represented below, but the reader should keep in mind that it is rapidly changing.

Toxicity of Manufactured Nanomaterials

Complications of Assays

Traditionally, ecotoxicology assays follow standard protocols and involve a group of species that has been selected to be representative of various organisms in the environment including Daphnia, bacteria, fish, birds, and insects. The most standard assays involve 24- to 48-hour exposures and mortality endpoints such as LD50. Other standard assays for these species include month-long trials assessing growth and reproduction, tumor development, and other endpoints. Some issues arise in conducting these assays for nanomaterials, which is one of the reasons the current toxicological information regarding nanomaterials is variable. Some of the major issues in toxicology assays include delivery of nanomaterials in media and approximating environmental conditions, characterizing exposures, maintaining exposures throughout an assay, and determining the state of exposure throughout an assay. For example, many nanomaterials are not easily dispersible and aggregate substantially when introduced into common exposure media. A hydrophobic particle will aggregate with other particles in water and to an even greater extent when salts are introduced into the media as is the case with many aquatic systems. This causes several issues when determining toxicity. First, the nanomaterial aggregate may no longer be in the nanosize range (hydrodynamic diameter >100 nm). In some cases this larger aggregate is nevertheless more toxic than its bulk chemical counterpart. This has not been shown with all nanomaterials, however, and is controversial. Second, the nanomaterials as they aggregate settle out of suspension, so depending on the organism involved the actual exposure may change over time as particles are effectively removed from suspension, or they may increase as some organisms (Daphnia) collect particles not only from the water column but from the bottom of the vessel. Some researchers have attempted to suspend the nanomaterials using solvents or detergents for better distribution. Studies have shown the solvent or detergent when not removed from the nanomaterials sufficiently can cause toxicity (Lovern and Klaper 2006; Henry et al., 2007; Kovochich et al., 2009) and in others the data indicates that this smaller aggregate size created by the additive changes the toxicity (eg, Lovern and Klaper 2006). Researchers have attempted to circumvent this issue by either changing the surface chemistry of the nanomaterial to make it more readily suspendable by adding a hydrophilic functional group or by altering the exposure conditions to include organic material that assists in suspension. Each of these approaches raises its own issues for toxicology. Changing the surface chemistry of a nanomaterial can also change its toxicity (Klaper et al., 2010). The relationship between surface chemistry and toxicity is important to understand because surface functionalization is an important aspect of nanomaterial formulation for industrial applications. So determining the impact of different surface chemistry changes is a necessary part of not only ecotoxicology but nanotoxicology in general. A bigger issue is that many coatings can cause toxicity on their own regardless of whether it is attached to the nanomaterial, and has also shown to be removed in certain cases by the organisms in the test media, or the environment the organism inhabits, which dramatically changes the properties of the nanomaterial through the experiment (Roberts et al., 2007). Is the toxicity then due to the original particle or now to the altered form of the nanomaterial?

As part of determining the dose an organism actually encounters is determining how much of any nanomaterial actually reaches the organism and is taken up into the organism and its tissues. There are several difficulties in measuring uptake including identifying the nanomaterial within the matrix of the organism versus inside the organism, for some nanomaterials the issue is distinguishing the nanomaterial from the matrix of the organism or environment either due to their small size relative to other aggregates in a sample, or because of the chemical similarity of the nanomaterial with the matrix (carbon-based nanomaterials in tissue, for example). Bulk extractions of these materials have provided some indication of potential concentrations (Westerhoff, Furgeson refs) but missing is the finer detail regarding location and structure of nanomaterials within a matrix, which is necessary to determine that the nanomaterial itself is causing an effect and not its byproduct or a secondary physiological response. Another issue in conducting a normal ecotoxicological assay using nanomaterials is the issue of calculating the dose as a mass versus surface area as reported above. Most nanomaterials when reaching the environment are not in a spherical uniform form, which complicates any calculations of dose based on surface area.

The real adverse impacts of nanomaterials may not be due to ambient environmental concentrations that arise but may be due to some subset of materials that are persistent and biomagnify in the environment. Research indicates that various nanomaterials are taken up by organisms in the environment and much is excreted by the organism within hours (Lovern et al., 2008; Petersen et al., 2009). In addition, gold nanomaterials and cadmium selenide quantum dots have been shown to accumulate in tissues of organisms and are able pass from one level to another in simple food chain experiments (Judy et al., 2011; Werlin et al., 2011). These include microbial food chains and those involving consumption of algae or bacteria into protozoa, herbivores, or detritivores. However, nanomaterials passing from one level to the next do not guarantee biomagnification up the entire food chain as seen in quantum dot studies using transfers from Daphnia to zebrafish or where protozoa take up quantum dots and accumulate them to a greater extent than the media concentration but rotifers feeding on the protozoa do not show further increases in concentrations of nanomaterials (Holbrook et al., 2008; Lewinski et al., 2011).

Ecotoxicity of Nanomaterials

The studies on the toxicity of ENMs to date have largely focused on simple structures of fullerenes, CNTs, metal oxides, and gold and silver particles with a few examples of the impacts of nanomaterials with varying surface chemistries or functional groups. The major conclusion from these data is that toxicity varies with the type of nanomaterial and is not universal across materials. There are varying degrees of toxicity depending on the type of nanomaterial, organism studied, and the co-occurrence of other environmental factors such as UV light, organic carbon, or low pH. The most common endpoints monitored in nanoecotoxicology are acute mortality measures with traditional toxicological models species such as Daphnia, zebrafish, trout, and Arabidopsis. Most studies find some degree of toxicity but the concentrations of most nanomaterials that are needed to kill half of the sample population are in the mg/L range, which is far above the estimates of potential exposures to any nanomaterial. Nanomaterials may enter other organisms just as they do humans and other mammals, through external contact with outer tissues and breathing structures (gill, skin), and through ingestion. Nanomaterials have been shown to be ingested with and without food present in most situations thus leading to contact with internal structures. The greatest damage measured has been to gill structures in fish and in digestive tissues in other species (Zhu et al., 2008; Bilberg et al., 2010). In terrestrial studies, plants have been shown to accumulate nanomaterials in root tissues and it is assumed that inhalation is possible in certain terrestrial situations but impacts on birds and mammals other than rats has not been demonstrated.

Fullerenes and CNTs were some of the first nanomaterials investigated for their potential environmental impact. Fullerenes have been shown to be lethal to Daphnia and microbial organisms (eg, Lovern and Klaper 2006; Zhu et al., 2006). However, ecotoxicity studies have indicated that the damaging impacts of these nanomaterials can be dramatically impacted by the method of introduction into exposure media. This includes introduction by solvent carriers or by altering the materials with surface chemistries that make them more water soluble. Unmodified fullerene nanomaterials may only be toxic at very high doses mg/L and mg/kg soil (Zhu et al., 2006; Li and Alvarez 2011; Mouchet et al., 2011) and have been shown to have little impact on bacterial communities and protozoa (Johansen et al., 2008). However, increasing their solubility through either solvent or detergent introduction or by changing surface chemistry can also increase their toxicity to aquatic species (Klaper et al., 2010).

Silver nanomaterials are some of the most widely used materials and appear to demonstrate the greatest toxicity of materials investigated in the literature. Silver in particular is toxic at ug/L doses to a variety of organisms (George et al., 2011; Bilberg et al., 2012). However, silver and other metal nanomaterials are subject to dissolution even with various coatings and their entire toxicity may be due to the ionic component of an exposure rather than the nanomaterial form (Kennedy et al., 2010). These metal ions may accumulate in the body of organisms after ingestion of nanomaterials and contribute to toxicity (Asharani et al., 2008, Li et al., 2011). Rather than creating a free radical in media the impacts of metal nanomaterials may be due to metal imbalance in cells after uptake and accumulation leading to apoptosis and cellular disregulation (Kao et al., 2012). However, it has been suggested that the dissolution and ionic form of the metals is not the sole reason for nanometal toxicities (Griffit et al., 2007). Nanosilver and possibly other nanomaterials based on soft metals may also react with environmental sulfides to produce silver sulfide nanomaterials, in which the silver bioavailability and toxicity is much reduced (Kim et al., 2010; Liu and Hurt 2010).

Mechanisms of Toxicity

As in mammalian toxicology studies, oxidative stress has been implicated as a major way in which nanomaterials exert toxicity either by generating free radicals within the suspension media or by changing the chemistry of the cells in which they come in contact. Metal oxide nanomaterials in particular have been found to generate oxidative stress. Hydroxyl radicals and oxidative stress have been found to increase over time in the presence of metal oxides and UV light (Yu et al., 2011). As in human tissues the stimulation of oxidative stress pathways alone are not an indication of harm and may be countered by adaptation by the organism to exposure. Mortimer et al. (2010) found that despite oxidative stress in initial assays, toxicity decreased over time in protozoa after exposure to metal oxide nanomaterials. However, Klaper et al. (2009) found that acute assays of oxidative stress do provide an early indicator of mortality of Daphnia in chronic assays for a variety of nanomaterial types. Metal oxide nanomaterials appear to have greater toxicity than their bulk counterparts. Titanium dioxide nanomaterials generate more ROS per gram than similar macroparticulates (Xiong et al., 2011). However, they do not differ from larger aggregates in their toxicity to developing fish (Zhu et al., 2008). Most studies have found a greater difference among nanomaterials composed of different metal oxides than the difference between the bulk and nanomaterial form of these compounds.

Metal nanomaterials have been found to cause a suite of effects, which in fish include negative impacts on respiration, oxidative stress, and development (Zhu et al., 2008; Shaw and Handy, 2011), in Caenorhabditis elegans increased mortality and decreased reproduction (Wang et al., 2012) and inhibit algal growth (Peng et al., 2011). Differences among metal oxides may be a result of differential solubilities in media.

The toxicity of nanomaterials to aquatic organisms can be greatly dependent upon the interaction of nanomaterials with the media to which they are introduced (Figs. 28-24 to 28-26). For example, dissolution of metal oxides and metal nanomaterials is greatly impacted by the characteristics of the media such as pH and salt content. Li et al. (2011) found that zinc oxide nanomaterials were most toxic in ultrapure water followed by salt, lithium borate (LB), and phosphate buffer saline (PBS), buffers. In the environment, toxicity may differ with the presence of organic matter and the specific interaction of organic matter with a nanomaterial will differ depending on its composition. The degree of interaction of nanomaterial and organic matter has been shown to impact toxicity (Gao et al., 2009). Toxicity of particles sorbed to natural organic matter can be significantly greater (Edgington et al., 2010) or diminished (Blinova et al., 2010). Toxicity of nanomaterials can be significantly impacted by the surface chemistry, capping agents, and coatings of the particles. Additions of functional groups have been shown to increase toxicity in some cases (Klaper et al., 2009; Klaper et al., 2010) and decrease toxicity in others (Allen et al., 2009). A few studies have also examined the influence of aging of nanomaterials in the environment on their toxicity.

Nanomaterials may also impact the bioavailability and toxicity of other contaminants in the environment. Fullerene and CNT particles in particular have been shown to bind to various organic contaminants with the strength of the association dependent upon the compound. Sorption to these nanomaterials can increase concentrations of bound organics by up to 60%, which if ingested could increase their toxicity (Baun et al., 2008).

As many nanomaterial exposures in the environment will occur at relatively low levels (ug to ng/L exposures), it is important to use assays other than LD50 to indicate the most relevant environmental effect. Including sublethal assays of physiology, assays of impacts of chronic low-level exposures, monitoring of alternative endpoints such as reproduction, growth, and at a higher level changes in community structure and ecosystem function are essential. Nanomaterials cause a shift in overall behavior in Daphnia, which can indicate a greater susceptibility to predators and a greater energy cost leading to decreased reproduction (Lovern et al., 2007; Braush et al., 2011). Various nanomaterials have been shown to impact development and physiology of fish species causing greater apoptosis in tissues, stress, and other impacts including structural effects (Chen et al., 2011; Shaw and Handy 2011; Truong et al., 2012; Xiong et al., 2011). As the immune response is the first interaction of foreign substances with an organism, it has been shown that nanomaterials are stimulatory to the immune system of fish in particular and have an effect that is equal to the response to bacterial cell components, which may indicate an eventual cost to the organism (Klaper et al., 2010; Jovanovic et al., 2011). Nanomaterials also have the potential to be mutagenic and in flies cause mutations that alter the phenotype significantly into the second generation (Vecchio et al., 2012).

Measurements of impacts at the community and ecosystem level are uncommon for nanomaterials, as they are for chemical toxicants with a much longer toxicology history. The existing studies indicate that community structure is impacted by the same nanomaterials that show toxicity and developmental impacts such as metals and metal oxides. Cerium dioxide and silver nanomaterials have a significant impact on marine microbial communities and microbial communities in wastewater—decrease in degradation potential and productivity (Fabrega et al., 2011; Garcia et al., 2012). They also cause shifts in decomposition communities in freshwater streams (Pradhan et al., 2011). Nanomaterials such as unfunctionalized fullerenes that have been shown to be less toxic exhibit no impact on soil community structure (Tong et al., 2007), macrophyte structure (Velzeboer et al., 2011), or mutagenesis across generations (Nyberg et al., 2008) but do cause shifts in community composition of soil bacteria (Johansen et al., 2008). The overall composition may also be altered as microbial communities on mucus secretions of carp are different in those that have been exposed to fullerenes (Letts et al., 2011). The overall community impact is greatly affected by complexity in environmental samples (Kang et al., 2009).

Ecosystem impacts may be impacted by these community effects as bacterial community function has a great impact on nutrient cycling. CNTs have been shown to impact microbial activity (Chung et al., 2011) and fullerenes, silver and CdSe quantum dots impact sediment bacterial oxidation of organics (Gao et al., 2009).