Certain organisms intentionally synthesize nanoparticles (eg, selenium-respiring bacteria, magnetotactic bacteria, gold nanoparticles in alfalfa).98,175,255 The diverse nature of the substances, compositions, structures, and physical properties involved in nanotechnology limits generalization regarding possible human effects. Representative of the knowledge gap, the US Hazardous Substances Data Bank identifies only few nanosubstances, namely, carbon black, carbon nanotubes, fullerenes, octinoxate, cerium oxide, iron, platinum, samarium oxide, silver, selenium, titanium oxide, and zinc oxide nanoparticles. Multiple reviews have attempted to address the diverse issues surrounding nanoparticle toxicity, and various schemes have been proposed to categorize nanoparticles.10,82,118,192,199,248,264,323 One older review reported 428 studies documenting adverse events of 965 unique nanoparticles.118 Reviews of a single “class” of nanoparticles have documented both adverse and neutral effects. Inference of systemic biological effects in humans is hindered by the in vitro nature of many experiments. Differences in experimental methodology, cell line, substance concentration, particle size and geometry, exposure parameters (route, duration, and frequency), duration of observation, and end points or surrogate markers have hindered comparisons. Data on the long-term effects are lacking. Cell-line specific, organ-specific, and species-specific toxicities remain to be fully characterized.
To address these deficiencies, methodologies are being developed to assess nanotechnology risk in a systematic fashion. Classifying nanoparticles based on the results of multi-dose-range biological assay profiles (ATP content, reducing equivalents, caspase-mediated apoptosis, and mitochondrial membrane potential) in multiple cell lines has been advocated.320 Other techniques to evaluate toxicity and biologic reactivity of nanoparticles have been reviewed and include cellular proliferation, necrosis, and apoptosis assays; reactive oxygen species (ROS) generation and oxidative stress; activation of proinflammatory signaling or other messenger molecules; genotoxicity and gene expression analysis; and in vivo exposure route, short and long term effects, tissue localization, biodistribution, and clearance studies.207,326,334 Studies and models of quantitative structure-activity relationships (QSAR) may take on more importance to profile toxicity as the number of compounds proliferates.286,381
Despite the many unknowns, antecedent work within the discipline of particle toxicology has provided extensive epidemiologic and experimental evidence associating airborne pollution particulate matter (PM) with human mortality, cardiovascular, pulmonary, neurologic, and reproductive injury, altered neurocardiac function (decrease heart rate variability and repolarization), and malignancy.36,67,71,153,268,272,297,366,403 The strength of the association depends on the particle size, type, and the outcome of interest. Similarly, ultrafine particles in home-generated cooking fumes have received attention for their possible role in pulmonary disease, inflammation, and genotoxicity.217,326,327,369,386
Genetic or unique susceptibilities to nanoparticles are not well categorized. Preexisting acute or chronic disease (pulmonary disease, cardiac disease, malignancy, infection) or individual genetic variations (resistance to oxidative stress, immune composition, surface or serum proteins) may modify nanoparticle toxicity. For example, following identical ultrafine particle aerosol exposure, patients with chronic obstructive pulmonary disease (COPD) received an increased “dose” (deposition factor and/or rate) compared to normal individuals.38 Retention of ultrafine particles was similarly higher in patients with COPD than healthy nonsmokers.224 Patients with preexisting cardiovascular disease or older age show increased susceptibility to concentrated ambient air pollution particles.67
Although the following sections generally report “positive” studies in order to highlight nanotoxicity principles, it is important to acknowledge that many other studies have also produced negative results, and some have challenged the concept of “nano-specific” toxicity.73 Toxicities identified experimentally must be coupled with an understanding of physicochemical properties (which impact organism and environmental transport) and exposure assessments in order to appropriately characterize risks.
Depending on the nanosubstance, “dose” (particle number or bulk amount) may or may not play a role. Appropriate dose-response curves (linear, supralinear, biphasic, or threshold) for different toxic effects have not yet been described for most particles.249 For example, anti-HER2 antibody-tagged silica-gold nanoshells showed no inherent in vitro toxicity to breast adenocarcinoma cells over a range of concentrations and exposure durations.189 CdSe/ZnS QDs encapsulated in phospholipid block–copolymer micelles injected into Xenopus blastospheres were “nontoxic” until doses of 5 × 109 nanocrystals/cell produced “abnormalities.”76 CdSe/ZnS QDs at higher doses (20 nM/L) increased apoptotic cell death and cytokine release.229 The principal toxic mechanism changed from intracellular oxidative stress to cadmium ion release as QD concentration increased in another study.182 Dose-dependent cytotoxicity was seen in human keratinocytes as SWCNT concentration increased from 0.11 to 10 μg/mL.204 Dose- and time-dependent effects were apparent in human peripheral blood lymphocytes exposed to oxidized or pristine MWCNTs.30 Nano-SiO2 caused cytotoxicity and induced the apoptotic pathway in dose- and time-dependent manners.398 In zebrafish embryos, silver nanoparticles induced dose-dependent embryotoxicity and multiple developmental abnormalities.178 Silver nanowires show dose-dependent cytotoxicity from 1.9 to 1900 μg/mL in human laryngeal epithelial and cervical carcinoma cells.1 Carbon black and fullerene (C60) manufactured nanoparticles were genotoxic in vitro and in vivo in a dose-dependent manner.353
Despite a comparable exposure on a mass-for-mass basis, nanoparticle toxicity may diverge significantly from bulk material. This may occur due to several factors including increased surface area, chemical reactivity, or ionization; altered absorption profiles; altered cellular interactions; or access to “protected” intracellular spaces. Cobalt-ferrite particles (6 nm) were more cytotoxic and genotoxic than 10 or 120 μm particles.55 Cytotoxicity of amorphous silica particles increased 33-fold as particle size decreased from 104 nm to 14 nm.235 Elemental carbon particles (5–10 nm) induced significantly more inflammatory mediators than larger carbon black particles (14 or 51 nm).16 CoCr nanoparticles generate more superoxide and hydroxyl free radicals and more DNA damage than CoCr microparticles, possibly due to faster dissolving or corrosion within the cell.262 TiO2 nanotubes (diameter 15–30 nm) permitted the best cellular adhesion, migration, viability, and differentiation compared to larger 50 nm nanotubes.266 CdTe QDs of approximately 2, 4, and 6 nm had over a six-fold difference in inhibition of human hepatoma cell growth.405 CdTe QDs (2.1 nm) can rapidly enter the nucleus of human macrophages and gold nanoparticles (1 nm) can penetrate cell and nuclear membranes.231,356 CdTe QDs (2.2 nm) were more cytotoxic than equally charged, larger QDs (5.2 nm).193 Nano-tungsten carbide-metallic cobalt particles (~ 80 nm) generated more hydroxyl radicals, induced greater oxidative stress, and caused faster cell growth/proliferation than fine (4 μm) particles.68
Toxicity may be primarily related to that of the bulk material. Elemental carbon particles (90 nm) were found to be significantly more toxic than diesel exhaust particles of comparable size (120 nm).320 Redox-active transition metals may pose a particular hazard.73 Gold, chrysotile asbestos, Al2O3, Fe2O3, ZrO2, and TiO2 nanoparticles differed in cytotoxicity in murine lung macrophage cells.330 QD core materials (cadmium, lead, selenium) can be toxic at relatively low concentrations to the plasma membrane, mitochondrion, and nucleus.52,120,194 Degradation of coated nanoparticles in the acidic and oxidative conditions of endosomes and lysosomes may result in exposure to inherently toxic core materials or ions. Derivitization or degradation may mitigate or exacerbate toxicity. Functionalization with quaternary amines prevented silica mesoporous nanomaterial-induced cellular injury.342 Conversely, compared to “pristine” SWCNTs, acid-functionalized SWCNTs blocked cell cycling of murine lung epithelium cells, and produced a more pronounce inflammatory response in mouse lungs in vivo.309 Similarly, oxidized MWCNT were significantly more toxic to human T cells than their “pristine” counterparts or carbon black.30 Decay of certain water-soluble fullerenes produces daughter compounds with increased toxicity in vivo.20
Contaminants may introduce or mitigate toxicity. “Doping” intentionally introduces impurities in order to modify the behavior of materials (eg, electrical properties of semiconductors). Similarly, doping can change the electronic, optical, and magnetic properties of nanocrystals.246 Nitrogen doping of MWCNTs improved biocompatibility and reduced lethality in mice exposed via multiple routes.43 Nitrogen-doped MWCNTs also proved more biocompatible in E. histolytica.81
The manufacturing process may unintentionally instill contamination—atomic or molecular impurities in the nanomaterial structure itself, residual reagents or catalysts, or manufacturing byproducts. Even “purified” SWCNTs retain significant percentages of cobalt, mobolybdenum, iron, nickel, yttrium, and zinc.170,203 Consequentially, non-purified, iron-rich SWCNTs (26 wt.% of iron) generated hydroxyl radicals and lipid hydroperoxides and depleted GSH more than “purified” SWCNTs (0.23 wt.% of iron).144 Purified fullerenes generate significantly less biological oxidative damage and adverse mitochondrial effects than unrefined fullerenes.17,285 Residual contaminants and impurities of substances used in surface modification of QDs are cytotoxic and genotoxic in vitro.130 Washing off unbound cetyltrimethylammonium bromide eliminated near total lethality of gold nanoparticles modified with this compound.56 Single walled carbon nanohorns, prepared from pure graphite without a metal catalyst, showed no skin irritation, eye irritation, or perioral toxicity.218
Coating and Surfactant Materials
Coating materials may shield toxicity of core compounds or possess inherent toxicity. They can mediate biocompatibility, duration of circulation, and organ- or cell-specific uptake. For example, various densities of electrostatic poly(glutamic acid)-based peptide coatings altered delivery of cationic polymer-plasmid DNA nanoparticles to liver, spleen, and bone marrow.121 QDs coated with carboxylic acids, amines, or PEG increased uptake by human epidermal keratinocytes in that order, with carboxylic acid-coated with QDs demonstrating cytotoxicity 24 hours earlier.300 Nanoparticle cores of 2-diethylamino ethyl methacrylate polymerized with poly(ethylene glycol) dimethacrylate were significantly more toxic to dendritic cells than cores surrounded by a 2-aminoethyl methacrylate shell.132 Upon inhalation, SiO2-coated rutile TiO2 nanoparticles but not uncoated rutile or anatase or nanosized SiO2 induced pulmonary neutrophilia and inflammatory markers.295
Conversely, covalent materials may reduce toxicity. Carboxylic acid grafting of cobalt-ferrite nanoparticles reduced toxicity by reducing leaching of Co2+ into solution.55 Gelatin coating reduced (but did not eliminate) cytotoxicity of CdTe QDs in human acute monocytic leukemia cells.40 PEG substituted CdSe core/CdS shell QDs were less cytotoxic extracellularly than “bare” QDs.45 However, some benign coatings can be rendered cytotoxic by air exposure or photodecomposition.65
Nanoparticles tend to aggregate due to attractive forces (van der Waals), and become progressively difficult to re-disperse as size decreases.280 SWCNTs agglomerations induced significantly more adverse effects than identical, well-dispersed SWCNTs.379 As CNTs are intensely hydrophobic, solvents or surfactant materials are used to disperse them and avoid clumping, although these materials themselves may be cytotoxic.379 Exposure to protein rich biological fluids may change the tendency to agglomerate, and therefore produce size-dependent effects. Using surfactants to disperse nanoparticles (eg, SWCNTs) can decrease protein adherence, and therefore alter biological effect or fate.78,103
Geometry and Architecture
Nanoparticle geometry may modulate toxicity and biological interactions. Previous observations suggested that toxicity might vary with the type of crystalline structure of a given material. For example, asbestos particle shape affects genotoxicity, and inflammatory and mutagenic properties vary by silica type (crystalline or amorphous).311 Identical concentrations of various aluminosilicate zeolite crystals (erionite, mordenite, and synthetic zeolite Y) of 0.1 to 10 micron size produced variant cytotoxicity and hydroxyl radical generation in rat lung macrophages.86 Material surface defects are important for ROS generation: amorphous TiO2 crystals were found to generate significantly more ROS than anatase, mixed anatase/rutile, or rutile crystals.140 In a separate analysis, nano-TiO2 (anatase) generates more biological oxidative damage than nano-TiO2 (rutile),17 and generally anatase nano-TiO2 is more photocatalytic than the rutile form.359
High aspect ratio is thought to contribute to nanoparticle pulmonary toxicity, including penetration of the alveolar wall and visceral pleural.73,215 SWCNT produced pulmonary granulomata and inflammation, whereas nanoparticle carbon black did not.17,203,325 “Long” (825 nm) CNTs, which were less easily enveloped by macrophages than their 220-nm counterparts, increased the degree of inflammatory response in rats.307 Asbestoslike pathology was also seen for “long” MWCNT (nanometer diameter, 15+ μm length) in mice.276 In mice, single wall nanohorns showed significantly less toxicity than SWCNTs.197 Nanofibers and graphene nanoplates with high aspect ratios but low aerodynamic diameter can persist to frustrate macrophage engulfment and contribute to toxicity.312 However, gold nanorods had significantly less cellular uptake than comparable spherical structures.50 This geometry dependence was reversed upon PEGylation.202 Attachment by HepG2 cells to micelles with PEG copolymer films varied depending on whether PEG polymer was in a “brush” (anchored at one end) or “mushroom” (anchored at both ends) conformation.132 Compared to control wafers of the same composition, Fe-Co-Ni nanowires hindered macrophage growth and development.3 Dendritic clusters consisting of aggregated 60 nm nickel nanoparticles produced higher embryonic toxicity than spherical 30-, 60-, and 100-nm particles of the same material.136
Additionally, nanoparticles may adsorb varying layers of surrounding biological proteins or lipids, altering effective size, shape, and density. Curvature and size of the nanoparticle may affect the extent of this “corona” and influence interactions with bound proteins such as albumin, apolipoprotein, and complement which may effect cells entry or receptor interactions.196,310 Bioassociation may also change nanomaterial properties. ZnO nanoparticles changed particle size, distribution, and charge in the cell culture media.375 SWCNTs adsorbed with serum proteins (primarily albumin) had an antiinflammatory effect, which was lost when surfactant-treated SWCNTs precluded adsorption.78 The same authors found, in contrast, prevention of protein adsorption to amorphous silica particles reduced toxicity.
Surface charge (zeta potential, ζP) may significantly alter the physical characteristics of nanoparticles, biological interactions, or effects. Neutral SWCNTs aggregate in aqueous solution; introducing a strong negative charge induces dispersal.309 Negatively charged nanoparticles permeated model pig skin, which excluded positively charged and neutral particles.160 Increasing surface charge density alters protein absorption; a positive charge promotes electrostatic association with negatively charged serum proteins.102,121 Charged nanoparticles are recognized as important inducers of complement activation.69 Strongly charged particles can also mediate direct membrane damage (hole formation) in the lipid bilayer.128,129 Positively charged polystyrene nanospheres induced oxidative stress, those with neutral charge did not.389 Toxic effects of strongly negatively charged acid–functionalized SWCNTs were abrogated by pretreatment with neutrality-inducing L-lysine.309 Positive charge mediates actin-dependent amorphous silica nanoparticle movement along filopodia and microvillilike structures.257 Mouse peritoneal macrophages and human hematopoietic monocytic cells show charge-dependent endocytosis—the higher the negative or positive surface charge of albumin particles, the greater the uptake.294 An overlap of the conduction band energy (Ec) level of metal oxide nanoparticles with the cellular redox potential (–4.12 to –4.84 eV)—which signifies the permissibility of electron transfers in biological environments—correlated with induction of oxygen radicals, oxidative stress, and inflammation.404 Also, within lysosomes, positively charged nanoparticles are capable of acting as “proton sponges,” enhancing cytoplasmic delivery and inducing cell death signaling.317
pH might be expected to effect toxicity by altering charge, solubility, protective or functional groups, bioavailability, and other mechanisms. pH alters the zeta potential of both microemulsion (3–5 nm) and hydrothermal (8–10 nm) cerium oxide nanoparticles, which in turn alters protein adsorption and cellular uptake.267 The significant toxicological difference in mice between 23.5 nm and 17 μm copper particles was judged secondary to stomach retention, with persistent depletion of H+ ions leading to systemic metabolic alkalosis and generation of ionic copper.214 CdSe core QDs exposed to low-pH simulated gastric fluid decreased cell viability in vitro, possibly due to degradation of the ZnS shell and increased solubility of cadmium from shell-free particles.373 Acidic intracellular organelle localization raises degradation and damage concerns; the inflammatory potential of 15 metal/metal oxide nanoparticle correlated with acidic conditions, hypothesized to be related to elimination of the corona in lysosomes to expose a charged surface.51 Targeted drug delivery employs intentional engineering of pH-responsive nanoparticles. In the acidic environment of the endosome or lysosome (~ pH 4.5), core-shell particles can swell by almost 3 times, disrupting these structures and allowing cytosolic entry.132 Alternatively, at altered pH, certain functional groups can be cleaved, resulting in altered charge and improved QDs delivery.221
At the cellular level, nanoparticle toxicity has been attributed to multiple mechanisms (Fig. 129–3). Oxidative stress, membrane and cytoskeleton structural alterations, cytoplasmic and nuclear protein interactions, energy failure, photoxicity, and genotoxicity have all been described. Nanoparticles may stress cell types differently. This may be due to engineered properties (associated targeting molecules), properties of nanomaterials themselves (size-dependent reticuloendothelial system deposition), or biological processes of target cells (eg, phagocytosis ability, resistance to oxidative stress, cytoskeletal architecture, cell division propensity). Tests in immortalized cell lines may not reflect in vivo findings due to inherent differences in genomic stability or specifically selected traits.
Potential mechanisms of nanoparticle cellular toxicity. ΔΨm = mitochondrial membrane potential, DNAmt = mitochondrial DNA; His = histone; MPTP = mitochondrial permeability transition pore; RNS = reactive nitrogen species; ROS = reactive oxygen species.
Unintended cellular uptake may contribute to toxicity. Depending on the particle architecture, several mechanisms have been described, including phagocytosis/endocytosis (clathrin-mediated, sca-venger receptor-mediated, mannose receptor-mediated, Fcγ receptor-mediated, complement receptor-mediated), potocytosis (caveolin dependent), (macro)pinocytosis, and direct cytoplasmic entry.70,99,239,265,322,348,362 Various types of ammonium-, acetamido-, fluorescein isocyanate-, and methotrexate/fluorescein isocyanate-functionalized SWCNTs and MWCNTs were internalized by a wide range of cell lineages (mammalian, fungal, yeast, and bacterial), some of which lack phagocytic and endocytic capacity, and under conditions which block energy dependent pathways.162 Dextran-coated supermagnetic iron oxide (SPIO) nanoparticles are taken up by human monocyte-macrophages and concentrated into lysosomes in a non-saturable manner.228 Nonendocytic uptake (eg, 78-nm fluorescent polystyrene microspheres and 25-nm gold particles) provides nonmembrane-bound nanoparticles direct access to intracellular proteins, organelles, and DNA.100
Engineered nanoparticles, which are capable of endocytosis, would be expected to translocate to distant sites within the body.82
Oxidative stress with subsequent lipid peroxidation, DNA damage, and apoptotic or necrotic pathway induction is a significant concern. Biologic oxidative damage as assessed by the ferric reducing ability of human serum revealed a diverse capacity for oxidative potential among nanoparticles.17 Metal oxide nanocompounds (eg, Co3O4, Cr2O3, Ni2O3, Mn2O3, and CoO) capable of electron transfers effectively produce oxygen radicals and oxidative stress.404 Research findings are complicated by differences in generation of reactive oxygen species (ROS) under abiotic conditions versus intracellularly.389 Biological mediums may also alter ROS generation.92 In neuroblastoma cells CdTe QDs induced lipid peroxidation and Fas cell death receptor upregulation, which could be mitigated by capping QDs with N-acetylcysteine.52 Adult zebrafish raised in solutions containing silver nanoparticles demonstrated oxidative injury, apoptosis, and cellular stress response (eg, GSH and p53 induction).54 Fullerenes have shown more conflicting results. Depending on experimental conditions and purity, they may either induce O2•– and OH• or possess antioxidant (and antiperoxidation) activity.104,392 SWCNTs and tungsten carbide-cobalt nanoparticles generate ROS and activate pathways leading involved in cytotoxicity and neoplastic transformation.68,204
Organelle and Substructure Damage.
Nanoparticles may compromise diverse subcellular components. Mitochondrial reduction capacity (eg, MTT assay) is one of the standard in vitro methods to assess nanoparticle toxicity. Nanoparticles’ mitochondrial toxicity includes alterations in mitochondrial calcium levels, dissipation of the mitochondrial membrane potential, lipid membrane destruction, and localization within the mitochondria.52,194,260,389 Small gold nanoparticles (3 nm) can access the voltage-dependent anion channel (porin) to cross the mitochondrial membrane.301 Induction of the mitochondrial base excision repair pathway enzymes suggests that MWCNTs can induce mitochondrial DNA damage.406 Polyamidoamine (PAMAM) dendrimers (45 nm) can down-regulate mitochondrial DNA-encoded genes involved in the maintenance of mitochondrial membrane potential and caused the release of cytochrome C, triggering apoptosis.177
QDs (2.1 nm) have been visualized to localize rapidly (< 30 minutes) and preferentially in the nucleus of human macrophages, mediated by endosomal transport along microtubular tracks.231 Co3O4 nanoparticles (45 nm) readily enter cells and their nucleus in vitro.263 SWCNTs can accumulate in the cell nucleus by crossing the lipid bilayer.277 Genotoxic consequences of this migration are described later.
Nanoparticles may cause disruption and compromise integrity of biological membranes by such means as hole formation in the lipid bilayer.128,129 Alteration in cellular adherence, migration, and actin cytoskeleton or microtubule function and structure have been described following exposure to SWCNTs, gold/citrate, TiO2, and SPIO nanoparticles.95,106,113,145,270 Calcium-mediated cytoskeletal function (as evidenced by cell stiffening) was altered by ultrafine carbon particles (12 nm, 90 nm) but not by diesel exhaust or urban dust particles.223 Functional processes may be impaired in the absence of the loss of viability. SiO2, TiO2, Ag, and Au, alone or functionalized with either positive or negative side chains, altered exocitosis and decreased secretion of chemical messenger molecules.191,209 Other protein functions may also be altered by nanoparticles. Copper nanoparticle clusters selectively induced unfolding and precipitation of hemoglobin A0 and E, whereas almost none occurred with hemoglobin A2.24 Nanoparticulate anatase TiO2 (5 nm) can directly bind lactate dehydrogenase and induced protein unfolding.75 C60 fullerene noncompetitively inhibits glutathione peroxidase in a substrate-specific manner.137 While relatively high concentrations were required (250 μM), the C60 fullerene derivative, dendrofullerene, selectively inhibited P450 metabolism of progesterone.91 Modification of protein structures by reactive nanoparticles might also serve as a mechanism for haptene formation and immunoreactivity.
Gene Expression and Genotoxicity.
Overwhelming nanoparticle toxicity can induce a cascade of expression of genes involved in either apoptotic or necrotic pathways. Nanoparticles can also upregulate stress- and inflammation-related genes as well as alter expression of genes involved in the cytoskeleton, trafficking, protein degradation, metabolism, growth and division, and detoxification. As a consequence of direct nuclear entry, nanoparticles can aggregate within the nucleus, bind directly to DNA or chaperone proteins, or induce nuclear membrane damage. Gold clusters (1.4 nm) present in growth media can access and directly associate with the major grooves of DNA to induce complete cell death in multiple carcinoma cell lines at concentrations at which cisplatin is only 10% effective.260,356,382 Nanosilver materials similarly bind with E. coli genomic DNA.396 Some QDs can target histones.231 Other nanoparticles (eg, SiO2) can cause clustering of critical enzymes such as topoisomerase I and sequestration of nuclear proteins (histones, splicing factor SC35, nucleolar protein fibrillarin, promyelocytic leukemia body protein PML, and p80 coilin), altering subnuclear architecture.48 DNA damage also occurs through induction of ROS.222
Nanoparticle damage may be addressed by base repair and other restorative mechanisms or may be so severe that DNA strand breaks, genome duplication or deletion, chromosomal instability, aneuploidy, malignant transformation, and proliferation occur in vitro and in vivo.133,355,406 Poorly tumorigenic or benign cells can acquire aggressive metastatic capacity when exposed to TiO2 nanoparticles.254 Alternatively, premature senescence was recently described using a C60 carboxylated adduct. Such an ability could be protective as a tumor suppression mechanism or compromise organ systems function.96
Nanoparticles containing photo-reactive dyes or intrinsic photothermal properties are useful for imaging and attractive for targeted chemotherapy. The approved liposomal verteporfin (benzoporphyrin derivative monoacid ring A) generates 1O2 following activation by low-intensity nonthermal laser light (689 nm). This is believed to induce cell death and prevent the loss of visual acuity in patients with subfoveal choroidal neovascularisation.151 Similarly, dextran-iron oxide nanoparticles linked to a photosensitizer which generates singlet oxygen caused complete cell death when exposed to specific light wavelengths.213 Nonmetal graphene QDs irradiated with blue light (470 nm) generated reactive oxygen species, including singlet oxygen, and killed human glioma cells by causing oxidative stress mechanism.206 C60(OH)22–26, a water-soluble C60 derivative under investigation as an antitumor, antibiotic, and drug delivery compound, was only mildly cytotoxic in the dark, but was significantly phototoxic to human lens epithelial cells via both type 1 (free radical) and type 2 (singlet oxygen) mechanisms under visible light.380 The concern is that ambient electromagnetic radiation could excite nontherapeutic nanoparticles to generate either ROS or thermal injury. Photolytic conditions have rendered coatings unstable, exposing toxic core metal components.120 Irradiation of TiO2, which can act as a photocatalyst, yields both oxidation and reduction reactions.94 Gold/citrate nanoparticles (13 nm) were capable of easily crossing human dermal fibroblast membranes and absorbing UV radiation.270 C60 fullerenes can generate ROS in the presence of visible light under physiologic conditions.393 The ROS extent depended upon C60 aggregation and associated stabilizing molecules in the aqueous phase.176
A host of miscellaneous cellular effects are still being characterized. Alterations in cellular resting membrane potential and ion channel functioning by metallic and carbon-based nanoparticles are a particular concern in neuronal tissue.18,185 Silver nanoparticles caused conformational changes of neuronal voltage-activated sodium currents and produced a hyperpolarizing shift and delayed recovery from inactivation.186 QDs can provoke intracellular lipid droplet formation due to accumulation of newly synthesized lipids and down-regulation of the β-oxidation of fatty acids.284 SPIOs can generate intracellular gas vesicles.201
A significant concern is the ability of nanoparticles to access “protected” spaces, such as the brain, the eye, and the reproductive tract. While targeted CNS drug delivery and imaging using engineered nanoparticles via intranasal or intravenous administration shows promise,88,150 unintended access is worrisome.
Seventy years ago, intranasal inoculation of small viral (nano)particles (the poliomyelitis virus, 25–30 nm) were shown to produce extensive polio invasion of the olfactory bulbs and lesions extending in a continuous series to the brain stem and beyond.26,27,131,354 Other small viruses (vesicular stomatitis, herpes simplex, and rabies virus) also access retrograde transport from the olfactory bulb to reach deeper brain structures.290 Anthropomorphic and engineered nanoparticles also may utilize this mechanism, with variable efficacy. Nanogold particles injected into rabbit olfactory areas demonstrated a direct connection between the olfactory bulb and the CSF.58 Intranasal instillation of SiO2-nanoparticles in rats led to wide distribution, striatal deposition, and a reduction in dopamine activity.385 In inhalation-exposed rats, ultrafine elemental 13C particles translocated into axons of the olfactory nerve.250 Rats exposed to aerosolized gold nanoparticles (20 nm) experienced particle distribution to the olfactory bulb and entorhinal cortex.402 Rats exposed to poorly soluble manganese oxide ultrafine particles (30 nm) showed olfactory bulb uptake and CNS delivery to the striatum, frontal cortex, and cerebellum.79 Other animal models demonstrated that cadmium, cobalt, manganese, mercury, nickel, and zinc reach the olfactory bulb when applied intranasally, and neuronal connections carry cobalt, manganese and zinc into deeper brain structures.35,123,271,352 Children and dogs exposed to pollution show prefrontal white matter hyper-intense lesions, associated with significant cognitive deficits in children. Ultrafine particulate matter deposition, vascular pathology, and neuroinflammation were evident in the dogs.41 Chronic respiratory tract inflammation might further diminish nasal respiratory and olfactory barriers and contribute to brain inflammation.273
Circulating nanoparticles may reach the brain. The blood–brain barrier (BBB) might be infiltrated via low density lipoprotein (LDL)–receptor mediated transcytosis of nanoparticles with nonspecific apolipoprotein adherence, endocytic processes, paracellular aqueous diffusion, or transcellular lipophilic diffusion.88,316 Intravenously administered water-soluble fullerene and intra-abdominally injected nano-TiO2 (5 nm, anatase) migrate to brain in experimental models.88,391 Nanoparticle entry can produce CNS inflammatory changes, inflammatory gene expression, demyelination, oxidative stress, lipid peroxidation, NO generation, and altered mitochondrial energy production.42,79,188,198,318 Cyclooxygenase-2, interleukin-1β, and CD14 upregulation in the olfactory bulb, frontal cortex, substantia nigrae, and vagus nerves and disruption of the BBB associated with particulate matter deposition were reported in patients exposed to high chronic pollution levels.42 Similarly, intravenous and intraperitoneal administration in rats of engineered silver, copper, and aluminum nanoparticles disrupted the ventral brain and proximal frontal cortex BBB to produce brain edema.318
Nanoparticles can contribute to secondary excitatory neurotoxicity. Intranasal delivery of carbon black (14 nm) increased olfactory bulb excitatory glutamate levels.351 Similarly, nano-Ag increased spontaneous excitatory postsynaptic currents and network activity.187 Given the number of CNS and systemic human amyloidoses, a concerning finding was that polymer nanoparticles, quantum dots, carbon nanotubes, and cerium nanoparticles all greatly enhanced the rate of ß2-microglobulin amyloid fibrillation by decreasing lag time for nucleation.184 Chronic dietary exposure to TiO2 nanoparticles in juvenile rainbow trout led to brain accumulation and 50% inhibition of Na+-K+-ATPase activity, which did not recover following removal from exposure.288
The lungs are expected to be the major portal of entry for anthropomorphic nanoparticles and engineered nanoparticles in manufacturing or research settings. Concerns include nanoparticle accumulation and persistent, acute and chronic inflammation, and surfactant disruption. In the bronchial airways, 24 hour retention depends on size fraction, which is negligible for particles greater than 6 μm but increases to 80% at 30 nm.165 In vitro studies show a variety of effects depending on the particle and experimental model. Canine and human alveolar macrophages uptake of elemental carbon particles (5–10 nm) induced lipid mediators AA, PGE2, LTB4, and 8-isoprostane in a dose-dependent fashion.16 SWCNTs activated alveolar macrophages, SWCNTs and C60 fullerenes disrupted plasma membranes, and SWCNT and MWCNTs effected antigen processing, presentation, and activation of T lymphocytes.116 In rat alveolar type 2 and epithelial cells as well as in human bronchial epithelial cells nanoparticulate carbon black particles (14 nm) induced dose-dependent proliferation via EGF-R and β1-integrin membrane receptors, phosphoinositide 3-kinases, and the protein kinase B (Akt) signaling cascade.363 Cerium oxide nanoparticles can be taken up by human fibroblasts.183 However, 3 hour exposures to aerosols of this fuel additive, while increasing catalase activity and minimally decreasing glutathione, did not alter cell viability, ATP content, TNF -α production, glutathione peroxidase activity, or superoxide dismutase activity in rat lung slices.87 Combustion-derived nanoparticles of organic compounds (1–3 nm) were mutagenic in Ames tests.315 Positive evidence for carcinogenicity of inhaled nanoparticles has been summarized.293
All types of carbon nanotubes and nanofibers are considered a respiratory hazard by the National Institute for Occupational Safety and Health (NIOSH).238 Carbon nanotubes can produce a variety of adverse effects in animal models: upper airway mechanical blockage, occlusive airway granulomatas, macrophage uptake and inter-macrophage carbon bridges, abnormal macrophage mitoses, type I pulmonary epithelial cell damage, lymphocytic proliferation, multifocal granulomatas, aggregation in alveolar spaces and interstitium, increased inflammatory and oxidative stress biomarkers, fibrinogenic reactions, alveolar wall thickening, bronchiolar epithelial cell hypertrophy, and peribronchial inflammation.143,170,203,226,238,324,374 At high doses, mice exposed to MWNTs orally, intraperitoneally, or via nasal installation showed essentially no tissue response, whereas intratracheal administration showed dose-dependent pulmonary tissue invasion, inflammation and granulomata formation, and death.43 Mice aspirating 40 μg of SWCNT and acid-functionalized SWCNTs had significantly higher BAL cell counts, PMNs, and cytokines (IL-6, TNF-α, and MIP2).309 Also of concern are MWCNT retention, persistent inflammation, and asbestoslike effects suggested by in vitro and in vivo experiments.275,298,330 A systematic review of 54 animal studies indicated that carbon nanotubes and nanofibers caused adverse pulmonary effects including inflammation (44 of 54 studies), granulomas (27 of 54), and pulmonary fibrosis (25 of 54), which were similar to other fibrogenic materials such as silica, asbestos, and ultrafine carbon black.238 Mice receiving both a known initiator chemical plus inhalation exposure to MWCNT were significantly more likely to develop tumors (90% incidence) and have more tumors than mice receiving the initiator chemical alone, suggesting that MWCNT can increase the risk of cancer in mice exposed to a known carcinogen.238
While rats exposed to C60 fullerenes nanoparticles (55 nm) for 10 days did not have visible of microscopic pathological lesions, particle burden and BAL protein concentrations were elevated.11 Intratracheally delivered single wall nanohorns generated no inflammatory response, although nanohorns accumulated and persisted.218 Intratracheal instillation of ultrafine (< 200 nm) particles from combusted coal induced a higher degree of neutrophil inflammation and cytokine levels than did the fine or coarse particles.109 Rats intratracheally instilled with ultrafine carbon black or TiO2 demonstrated neutrophil recruitment, type 2 epithelial cell damage, cytotoxicity, impaired macrophage phagocytic ability, and enhanced macrophage sensitivity to chemotactaxins.291 A summary of nanosized TiO2 particle effects concluded that the crystalline form has the greatest impact on pulmonary responses, whereas particle surface area, alumina or amorphous silica coating, and shape have lesser influences on toxicity.199 Nasal instillation resulted in neutrophil recruitment. Alveolar macrophages took up inhaled elemental silver nanoparticles, and deposition in the alveolar wall was noted.339 While particles were rapidly cleared from lungs, there was evidence of circulatory spread. Longer-term studies showed significant silver lung persistence, chronic macrophage accumulation and alveolar inflammation, and systemic distribution.337 The severe toxicity of air-generated polytetrafluorethylene (PTFE fumes) can be reduced by aging, filtering, and preexposure, suggesting nanoparticle upregulation of pulmonary inflammatory cytokines and antioxidants.142
Pulmonary surfactant provides low surface tension and can identify and bind targets for phagocytosis. Independent of cellular effects, in vitro lung models suggest that nanoparticle deposition may cause pulmonary surfactant dysfunction during the breathing cycle.148 The collectins (phagocytosis enhancers), human surfactant protein-A and -D, bind double wall carbon nanotubes in a calcium-dependent manner.303 Diesel exhaust particulate matter can be solubilized and dispersed in the major component of pulmonary surfactant (dipalmitoyl phosphatidyl choline) and induce genotoxicity in multiple different assays in bacteria and mammals.303,370
A report of uncontrolled workplace exposure to polyacrylic ester nanoparticles and other materials causing retained intracellular nanoparticles, pulmonary inflammation, pulmonary fibrosis, and pleural foreign-body granulomata was criticized for lack of causality and differential diagnosis.32,329 Similarly, a report of submesothelial deposition of carbon nanoparticles after toner exposure has been criticized for lack of plausibility and causality.347,378
Cardiovascular and Hematological Systems.
Ultrafine particles can alter human cardiac function (decreasing heart rate variability and repolarization).67,119,350,367,403 Heart rate variability provides a measure of cardiac autonomic control; a decrease predicts mortality in subjects with prior myocardial infarction.350 Significant pulmonary inflammation is not a prerequisite, although autonomic reflexes from pulmonary nerve endings may mediate these effects.105 Jugular vein or femoral artery administration of silver, copper, and aluminum nanoparticles immediately but transiently slows heart rate.318 Engineered nanoparticles made of flame soot (Printex 90), spark discharge generated soot, anatase TiO2, and SiO2 induced catecholamine-mediated dose dependent increases in heart rate and dysrhythmias in Langendorff heart model systems.331 Human wood smoke exposure increases levels of serum amyloid A (a cardiovascular risk factor) and factor VIII in plasma.15 Inhaled TiO2 nanoparticles impair rat coronary arteriole endothelium-dependent vasoreactivity and relaxation, likely due to microvascular ROS.173
Intravascularly, nanoparticle characteristics and concentration determine leukocyte and immune stimulation (detailed in a later section), erythrocyte effects, thrombogenesis, endothelial dysfunction, and atherogenesis. Dendrimers can adversely alter human erythrocyte morphology, induce clustering, and provoke significant hemolysis.72,77 Dendrimer generation, concentration, charge, exposure duration, and material type determine the extent to which this occurs. Intravenously provided diesel exhaust particles aggregate within erythrocytes, decreasing counts.243 Polycationic (but not neutral or anionic) water-soluble fullerene C60 derivatives, surfactant stabilized poly (lactic coglycolic) acid (PLGA) nanoparticles, and polystyrene nanoparticles all induce significant hemolysis.29,155,210 Chitosan/pDNA and uncoated polymer nanoparticles both induce significant hemoagglutination.121,190
Engineered and combustion-derived carbon nanoparticles (MWCNTs, SWCNTs, and mixed carbon nanoparticles) stimulated human platelet aggregation, activation of GPIIb/IIIa, and accelerated the rate of vascular thrombosis in rat carotid arteries.287 The effect on platelet aggregation was aspirin-resistant but was reduced by low affinity (P2Y12) ADP receptor antagonism. Multiple metal nanoparticles (iron, copper, gold, cadmium sulfide) also induced dose-dependent platelet aggregation through the P2Y12 ADP receptor; an effect was blocked by clopidogrel.62 Epidemiologic studies support an association between exposure to particulate matter of less than 10 microns and increased risk of deep venous thrombosis.9 Diesel exhaust particles (20–50 nm) and positively charged polystyrene nanoparticles cause rapid activation of circulating blood platelets and microcirculatory thrombi.240,242 Healthy volunteers exposed to concentrated ambient air particles increase fibrinogen levels.107
Compared to larger particles, ultrafine particles produce larger atherosclerotic lesions, decrease the antiinflammatory effect of HDL, and induce oxidative stress in susceptible animal models.6 At doses which produced no significant pulmonary inflammatory changes, rats exposed to nano-TiO2 aerosols display systemic impaired endothelium-dependent arteriolar dilation, outright constriction, and decreases in NO bioavailability by increasing local reactive oxidative and nitrosative species.247
Nanoparticle interaction with the immune system is complex: depending on biophysiochemical properties, nanoparticles can stimulate, silence, or elude immune responses.407 Most concerning is induction of chronic inflammation. Work in murine macrophage cells lines demonstrated that MWCNTs produced a cytotoxic response nearly identical to asbestos.330 Commercially and academically available MWCNTs induced asbestoslike, length-dependent pathology in mice.275,360 MWCNTs can reach the subpleura in mice after a single inhalation exposure to produce mononuclear cell aggregates and subpleural fibrosis.298 On the other hand, immune suppression may also occur. Fullerene suppression of mast cell–mediated hypersensitivity reactions appears to proceed from endocytosis, endoplasmic reticulum accumulation, intracellular persistence, and the inhibition of calcium and ROS generation.64
Particle-specific interactions may yield cellular accumulation, antigen mediated-immune stimulation, inflammatory mediator release, or fibrous or granulomata formation. Mouse granulocyte-macrophage colony formation is now a standard testing methodology for assessing nanoparticle toxicity.7 Macrophages exposed to Fe-Co-Ni nanowires in vitro increased production of IFN-γ, IL-1α, IL-4, and IL-10.3 Ultrafine carbon caused macrophages to release225 lipid mediators arachidonic acid, PGE2, LTB4, and 8-isoprostane.16 Subcutaneously implanted “hat-stacked” carbon nanofibers in rats generated fibrous connective tissue formation and macrophage recruitment and ingestion without inflammatory response (necrosis, degeneration, or neutrophil infiltration).400 Intratracheal exposure to nano-sized nickel and cobalt particles induced rat blood neutrophils to release reactive oxygen species and reactive nitrogen species.220 Human volunteer exposures to concentrated ambient particles (< 200 nm) increased inflammatory blood mediators.110 IV administration of ultrafine diesel exhaust particles promoted monocyte and granulocyte proliferation.243 Other cell lines can also be induced to release inflammatory mediators. SWCNTs induced human epidermal cells exposed to release interleukin-8. MWCNT introduced in human neonatal epidermal keratinocytes increased IL-8 and IL-1β release.382
Direct immune stimulation can also occur. Fullerenes can induce an IgG response capable of cross-reaction with other fullerenes.31,46 The degree of induced opsonization by various IgG and complement molecules is thought to explain part of the differences seen in blood clearance and tissue distribution of nanoparticles.259 Upon readministration, PEGylated liposomes have accelerated blood clearance due to IgM binding; this is inhibited by chemotherapeutics which inhibit B-cell proliferation.135 In the absence of direct immune stimulation, carbon nanotubes can boost immune response as adjuvants.261 Indeed, pharmaceutical nanoparticles show promises for eliciting polyclonal and monoclonal antibodies to nonimmunogenic haptens as diverse as herbicides, antibiotics, and vitamins.205
Liposome- and polymer-based therapeutic nanomedicines have been compromised by non-IgE mediated hypersensitivity reactions due to complement system activation which releases C3a and C5a anaphylatoxins.5 The drug solubilizer Cremophor EL (polyethoxylated castor oil) forms micelles (8–25 nm) at concentrations above 60 μg/mL and causes its occasionally severe immune reactions presumable due to complement activation. Carbon nanotubes can activate human complement via both classical and alternative pathways; C1 q binds directly to carbon nanotubes.302 Surface density of coating materials can also alter complement consumption.4 Certain dendrimers may also strongly activate the complement system.226 MnFe2O4 magnetic nanoparticles (10-nm diameter) induce severe inflammatory reactions in mice.169 Microcrystalline silica particles induced a strong Th1 response, whereas carbon black particles and polystyrene particles induced a mixed Th1:Th2 response.368 PLGA-based nanoparticles did not induce complement consumption, indicating that this action is nanoparticle specific.44
The ocular system is of particular interest due to its immunological privilege and physiologic characteristics. Therapeutically, nanosuspensions show promise for extended release and decreased drug clearance from the cornea and diminished systemic drug exposure.114,283 In vitro studies have shown adverse effects. Silver nanoparticle block the proliferation and migration and inhibit cell survival of bovine retinal endothelial cells.146 Water-soluble fullerene derivatives are phototoxic to human lens epithelial cells under visible light.380 Conversely graphene oxide nanosheets, even when intravitreally injected, did not demonstrate significant effects.395 Cerium oxide nanoparticles appear to offer a protective effect, scavenging reactive oxygen intermediates in rat retina cell culture.47 In vivo confocal neuroimaging conclusively demonstrated negatively charged nanoparticle-rhodamine formulations injected into rat jugular vein crossed the blood-retinal barrier within 40 minutes.282 Intravenously administered gold nanoparticles (20 nm) can bypass the blood-retinal barrier to distribute in all retinal layers within 24 hours.156
Reproduction and Development.
Reproductive effects of nanoparticles are incompletely characterized. One review concluded that nanoparticles can breach the blood-testis barrier and locate in the testes.212 Subsequent studies using inhaled fluorescent magnetic nanoparticles (50 nm) and injected gold nanoparticles confirmed testicular distribution and persistence.12,168 Chronic intratracheal administration of carbon black (14 and 56 nm) showed similar partial vacuolation of the seminiferous tubules and decreased daily sperm production.401 Silver nanoparticles caused dose-dependent mitochondrial impairment and cytotoxicity in spermatogonia in vitro; aluminum nanoparticles caused apoptosis.34 Gold nanoparticles can penetrate human sperm cells, resulting in fragmentation and dysmotility.383 Environmental exposure to relevant concentrations of ZnO, TiO2, SiO2 nanoparticles produced reproductive toxicity in nematodes, which was correlated with ROS production and attenuated with antioxidants (ascorbate and N-acetylcysteine).387
Female reproductive effects are less well described. Oral ZnO nanoparticles achieved lactation and placenta transport and resulted in increased post-implantation loss rate, decreased live births, and zinc accumulation in offspring liver and kidney.141 In perfused human placenta models, PEGylated gold nanoparticles (10–30 nm) were retained mainly in the trophoblastic cell layer and internalized by trophoblastic cells and did not cross in detectable amounts into the fetal circulation over 6 hours of evaluation.230 Aquatic studies have illustrated detrimental reproductive and developmental effects. Daphnia magna were unable to reproduce again after exposure during pregnancy to sub-lethal concentrations of water stable fullerenes. Less than 10% of daughter daphnids matured following maternal exposure.341 Following exposure to silver nanoparticles (0.1 or 0.5 mg/L for 3 days), nematode offspring decreased by 70%; oxidative stress was the presumed toxicological mechanism.292 In the animal model of organogenesis, medaka (rice fish) eggs easily took up 39.4-nm fluorescent particles and concentrated them in the yolk area and gallbladder during embryonic development.149 Adults demonstrated diffuse uptake into reproductive organs as well as the brain, liver, intestine, gills, and kidney. Medaka fertilized eggs exposed to silver nanoparticles produced a variety of malformations including systemic edema, hemostasis, and vertebral, finfold, optic, and cardiac abnormalities over the range of concentrations tested (100–1000 μg/L).388 Altered hormone levels and offspring sex ratios seen in some experimental models of nanoparticle exposure suggest endocrine disruption.195,233
Bioaccumulation and Persistence.
Nanoparticles might accumulate in cells, tissues, or organs to achieve threshold toxicities at a later time than initial exposure. At the cellular level, nanocrystals can be passed to daughter cells upon mammalian cell division.117,208 Retained fluorescing QDs have been used experimentally to trace Xenopus cells from embryo to tadpole stage.76 As with different fine particles (fiberglass, rock wool, slag wool, asbestos), their biopersistent potential seems to underlie toxic effects.125 This knowledge drives the concern regarding preliminary studies demonstrating MWCNT persistence, inflammatory induction, and asbestoslike pathology.225,275,298,338 Compared to sodium selenite, nano-selenium particles displayed hepatic hyperaccumulation and persistence in fish models, with associated oxidative stress and toxicity.181 Similarly, uptake, bioconcentration, and alimentary toxicity of TiO2 nanoparticles were demonstrated in crustaceans.59 Their transmission and bioaccumulation (by a factor of over 100% in certain species) up the food chain was also apparent.399 Nanocrystalline C60 had a greater propensity to accumulate in D. magna fetuses due to their higher lipid fraction (including the egg sac yolk) and correlated with higher mortality.341 Even short-term oral exposure may lead to reticuloendothelial accumulation in some animal models.343
While desirable for chemotherapeutic applications, prolonged periods of persistence could allow leaching and toxicity of initially protected materials.377 Bone marrow deposition of nanoparticles278 could allow for ongoing systemic exposure. Persistence in other compartments could also permit more extensive distribution (eg, translocation from lung tissue). Dextran-coated ultrasmall SPIO particles currently used in clinical trials are retained by human monocyte-macrophages for days.227,228 QDs persisted at least 4 months in mice, while those conjugated to bovine serum albumin or coated with mercaptoundecanoic acid underwent negligible elimination in rats.13,90 Due to intrinsic resistance to lysosomal degradation, inorganic or metal nanoparticles can have extremely low rates of excretion, resulting in long-term accumulation.377