Katrien
Luyts
,
Dorota
Napierska
,
Ben
Nemery
and
Peter H. M.
Hoet
*
Department of Public Health, K.U. Leuven, Herestraat 49 (O&N 1, bus 706) B-3000 Leuven, Belgium. E-mail: peter.hoet@med.kuleuven.be
First published on 23rd November 2012
The increased use of and interest in nanoparticles (NPs) have resulted in an enormous amount of NPs with different compositions and physico-chemical properties. These unique properties not only determine their utility for (bio-medical) applications, but also their toxicity. Recently, “nano-researchers” became aware of the importance of determining the characteristics since they might be predictors of their toxicity. Currently, we face a large set of (non-coordinated) experiments with miscellaneous objectives resulting in a large quantity of available (and often incomplete) data, which hamper the unraveling of the complex interrelated NP characteristics with experimental results. Here, we try to link different critical physico-chemical characteristics separately with toxicity observed in both in vitro and in vivo models.
Environmental impactRecently engineered nanomaterials can be found in a range of commercial products, and without doubt more and more nanoparticles (NPs) will enter our environment. Unique – size dependent – properties not only determine their utility, but also their toxicity/hazardous properties. While size – based on the existing definitions – seems to be the most important property of NPs, it is now clear that other physico-chemical characteristics are (equally) important – and many of these characteristics are interrelated. In this overview, we try to link different critical characteristics separately with toxicity observed – in both in vitro and in vivo models – in order to unravel the complexity of the nano-related hazardous effects. |
Oberdörster et al.2 proposed a list of physico-chemical characteristics that might be important to understand the biological activity and toxic properties of NPs. Particle size and size distribution, agglomeration state, shape, crystal structure, chemical composition, surface area, surface chemistry, surface charge and porosity were suggested as key characteristics.
On the other hand, the enormous diversity of NMs makes it very difficult to compare results of different studies when their characteristics are too diverse. The more recently published paper by Bouwmeester et al.3 presents the outcomes of a workshop of the European Network on the Health and Environmental Impact of Nanomaterials (NanoImpactNet). During the workshop, experts in the field of safety assessment of engineered nanomaterials (NMs) addressed the need to systematically study sets of engineered NMs with specific metrics to generate a data set which would allow the establishment of dose–response relations. The proposed list of minimal characteristics and metrics includes, among others, chemical composition, size distribution, surface area, structure, shape and persistence.
Despite the research performed over the past 10 years, there are still many gaps in the understanding and knowledge regarding nanotoxicology. This is, in part, caused by the complexity of nanotoxicological studies. It was quickly realized that the techniques that are suitable to study chemical toxicity could not always be applied in nanotoxicology, since NPs may behave very differently and can interfere with the toxicological assays themselves.4 Subsequently, nanotoxicological studies had to be designed differently which adds to the complexity of the “nano field”.
The unexpected toxicity of NPs is caused by the complexity of their interrelated physico-chemical properties but since the start of the nanotoxicological research5 there has been no clear understanding of the relations between NP toxicity and their physico-chemical properties (obviously, except for size). This is the rationale for this literature review where we attempt to link several NP properties separately to their toxicity.
This inverse relationship between size and toxicity can, in part, be explained by the fact that smaller particles have a higher ratio of surface to total atoms or molecules contained by the particle. Auffan et al.6 postulated that most types of particles have a critical size of about 30 nm below which NPs exhibit their typical “nano” properties different from their bulk material. From this size on, the ratio of surface to total atoms or molecules increases exponentially with decreasing size (Fig. 1).7 Below their critical size, NPs are characterized by an excess of energy at the surface and are, therefore, thermodynamically less stable resulting in increased surface reactivity. Consequently, a larger amount of surface molecules with a higher potency to react due to the thermodynamic instability are both in line with the increased toxicity.
![]() | ||
Fig. 1 The percentage of molecules localized at the surface of a nanoparticle, expressed as a function of the nanoparticle size. When the size of a nanoparticle decreases, the amount of molecules present at the particle's surface increases in an exponential trend. Adapted from ref. 7. |
In the following section we will describe, on the basis of a handful of studies, that, indeed, size matters.
Cho et al.16 and Hirn et al.17 studied the biodistribution of silica NPs in mice and gold NPs in rats respectively after intravenous (i.v.) injection. They both found accumulation of mainly the larger particles (Cho study: 100 and 200 nm; Hirn study: up to 99% for 200 nm) in the macrophages of the liver and spleen. The smallest NPs (Cho study: 50 nm; Hirn study: 1.4 nm) were cleared earlier via urine and bile compared to the larger particles. Hu et al.18 showed an increased uptake efficiency with smaller copolymer NPs (39.9 and 47.2 nm) in HepG2 (liver hepatocellular cells) compared with larger ones (149.1 nm) with the same copolymer configuration. After instillation of gold NPs in rat lungs, Semmler-Behnke et al.19 showed translocation of 1.4 nm gold NPs through the air–blood barrier of the respiratory tract whereas 18 nm particles were almost completely trapped in the lungs. Choi et al.20 investigated, in a rat model, the role of the diameters of instilled NPs (5–300 nm) in the translocation from the lungs to extrapulmonary compartments of the body. This study demonstrated that NPs with a hydrodynamic diameter smaller than 34 nm translocate rapidly from the lungs to mediastinal lymph nodes and NPs with a hydrodynamic diameter <6 nm can traffic rapidly from the lungs to lymph nodes and the bloodstream, to be subsequently cleared by the kidneys.
In conclusion, a growing number of studies show that smaller nanosized particles often cause more toxicity than larger particles. However, some conflicting data can be found, which sometimes can be explained mechanistically (e.g. membrane rupture of RBC) but often remains unexplained probably because unknown (or not measured) characteristics of the NPs remained uncovered.
This has considerable repercussions on the biokinetics of the material. Several questions can be raised: what is the size distribution of the aggregates/agglomerates; what is the portion of the particles present as a monodisperse material; does a size threshold exist allowing translocation through the body or can only monodisperse particles translocate between cells/organs? Aggregation/agglomeration influences the settling of inhaled materials in the different lung regions. Moreover, local effects are, in part, governed by the size since large aggregates cannot easily be removed (e.g. carbon nanotubes) from the initial site of deposition allowing prolonged exposure, possibly leading to chronic effects.
At the cellular level, active uptake of large aggregates – which may be larger than the cell volume – is often not possible, thus reducing bioavailability for most cell types, but initiating “frustrated phagocytosis” by specific cell types.
In addition to the issues of distribution and the bioavailability of the material, intrinsic toxicity can be affected by aggregation and aggregation size too. For many endpoints it is not yet clear whether the aggregation/agglomeration plays a major role – comparing equal levels of internalized/available material – in the toxicity of the material. In the case of amorphous nanosilica in hemolysis, this has been worked out in some detail. As mentioned earlier, the hemolytic activity of silica NPs increases with increasing size. Notwithstanding this observation, aggregates of “small” silica NPs are less potent than monodisperse materials, but the toxicity increased when the aggregates were denser and, hence, were displaying a smaller total surface area compared to the loose aggregates. These observations demonstrate the complexity concerning surface area, size and aggregations.
How shape or geometry governs cellular uptake is not always understood, but recently a few models describing the active cellular uptake of NPs have been published.
![]() | ||
Fig. 2 Local particle geometry determines phagocytosis by macrophages. (A) Spreading of the macrophage membrane around the surface of a disk-shaped NP. This process occurs in a tangential fashion where T is the average of tangential angles θ = 0 to θ = π/2. Ω represents the angle between N (the normal at the area of initial particle–cell contact) and T. (B) Internalization velocity (the course traveled by the membrane over internalization time) as a function of Ω. Each dot represents a different NP shape, size or aspect ratio and the bars represent SD. The arrows above the differently shaped NPs indicate the attachment site that corresponds to Ω on the x axis in the plot. When Ω ≤ 45°, the NPs are internalized, proven by the positive internalization velocity. Above this critical angle no internalization takes place and the cell membrane spreads over the particle. “Reprinted with permission from Champion & Mitragotri”.24 |
Chithrani et al.25 studied the shape (and size) effects of gold NPs on uptake into mammalian cells. They reported a higher uptake of 375% and 500% with spherical NPs (14 and 74 nm) compared to rod-shaped gold NPs (14 × 40 nm and 14 × 74 nm) respectively. They speculated that the rod-shaped NPs might have a larger contact area with the cell membrane receptors when interacting with their longitudinal axis compared to spherical NPs. Considering NPs with large contact areas, the local reduction in available cell membrane receptors might explain a reduced uptake since NP binding might occur less efficiently. This model only applies to (receptor-mediated) endocytosis, which implies that only particles with a minimal size at 50 nm undergo this process.
Both models described above focus on how NPs and cell membranes make contact, and how shape and/or volume can favor or hinder – at least in part – the internalization of NMs. However, applying these models to interpret the results of other available studies, with different NPs, coatings and cells, is most often not possible. Moreover, NPs can also be internalized by other mechanisms than endocytosis, although these mechanisms are less well understood. Thus, Geiser et al.26 showed in vitro that fluorescent polystyrene microspheres (1, 0.2, and 0.078 μm) were taken up by passive diffusion or adhesive properties (endocytic processes were excluded).
Gratton et al.29 found that monodisperse hydrogel nanorods with AR = 3 were taken up about 4 times faster by HeLa cells than their symmetrical counterparts of the same volume. Meng et al.30 showed that mesoporous silica nanorods with an AR of 2.1–2.5 were maximally taken up by HeLa and A549 cells compared with the same NPs with a lower (spheres, AR1) or higher (4.5) AR. Yu et al.31 concluded that mesoporous silica NPs with AR = 8 had a decreased hemolytic activity than silica NPs with a lower AR (2 and 4), whereas no correlation could be found between AR and cytotoxicity in A549 and RAW 264.7 cells. The study by Herd et al.32 also found no clear relationship between the silica NPs' shapes (spheres, worms and cylinders) and cytotoxicity in A549 and RAW 264.7 cells. Arnida et al.33 showed that polyethyleneglycol(PEG)-ylated gold nanorods (AR = 3.5 and 4.5) were taken up by prostate cancer cells to a larger extent than PEG-ylated gold nanospheres (AR = 1). Conversely, the studies performed by Hutter et al.34 and Chithrani et al.25,35 showed a larger uptake of gold nanospheres compared to other shapes by neuronal and HeLa cells, respectively. The study performed by Zhang et al.36 showed that spherical copolymer NPs were taken up to a larger extent by CHO cells than rod shaped NPs. Aaron et al.37 found that rod-shaped quantum dots diffuse into the plasma membrane and are internalized slower by immune cells compared to their spherical counterparts. Ji et al.122 showed that length (≥200 nm) and AR (≥22) were critical for CeO2 nanorods to induce lysosomal damage in THP-1 cells.
Although some inconsistencies appear, these studies indicate that the shape (geometry) of a NM–NP may in itself alter its biological properties in vitro. The inconsistencies may be explained by additional characteristics such as surface modifications (PEG vs. cetyl trimethylammonium bromide (CTAB) as gold NP stabilizer, peptide linking) and the cell types used (e.g. macrophages, RBCs, epithelial cells).
The few in vivo studies that have specifically investigated the effects of NP shape are more consistent. Huang et al.38 designed two differently shaped mesoporous silica NPs with similar particle diameters (≈70 nm), chemical compositions and surface charges but different ARs (1.5 and 5), and investigated the effects of NP shape on biodistribution, clearance and biocompatibility after i.v. injection in mice. Their results indicated that the low AR particles were trapped easily in the liver, whereas the higher AR particles appeared more in the spleen. The high AR NPs had a longer blood circulation time and were excreted by urine and feces less rapidly than the low AR NPs. Geng et al.39 i.v. injected mice with copolymeric nanofilaments (AR > 20) and showed that the long filaments had a circulation time about ten times longer than their spherical counterparts (diameter not mentioned). In vitro, under fluid flow conditions, spheres and the short nanofilaments were taken up to a larger extent by activated human-derived macrophages than the long filaments, which might explain the longer circulation times of higher aspect ratio NPs in vivo. Apparently, the longer filaments are not easily cleared and are, therefore, more biopersistent, possibly as a consequence of their impaired internalization, compared to spherical NPs.
A number of in vivo studies have investigated the pulmonary toxicity of SWCNTs and MWCNTs administered via pharyngeal aspiration or i.t. instillation in rodents. Results are consistent in showing acute inflammation, granuloma formation and, in some cases, the onset of fibrosis.41–47
The mouse peritoneal cavity has been used as a model of direct mesothelial exposure and a length-dependent response was shown, where longer fibers induced more extensive inflammation and granuloma formation within a short period after exposure.48–50 Muller et al.51 did not observe any carcinogenic response in rats, two years after a single administration of 0.7 μm long MWCNTs, although acute inflammation was present shortly after administration. Peritoneal tumors such as lipomas, liposarcomas and angiosarcomas were, on the other hand, present at terminal sacrifice.
When CNTs translocate to the pleural space, they might be cleared through stomata in the parietal pleura, which are pore-like structures linking the peritoneal cavity to the underlying lymphatic capillaries. When they are not cleared, CNTs might cause pleural pathology. Donaldson et al.52 hypothesized that the retention of long fibers at the stomatal openings on the parietal pleura, coupled with frustrated phagocytosis of pleural leukocytes that attempt to ingest them, produces a chronic pleural mesothelial inflammatory response. Chronic inflammation is known to be a driver for proliferation, genotoxicity and growth factor synthesis and release thus possibly resulting in conditions such as fibrosis, pleural effusion and mesothelioma. Al-Jamal et al.53 found that shortened, functionalized MWCNT–NH3+ were taken up by human pulmonary epithelial A549 cells and human primary macrophages. The internalization of MWCNT–NH3+ was the result of either membrane wrapping or direct membrane translocation of individual fibers or clusters of fibers within vesicular compartments.
Elgrabli et al.54 reported that MWCNTs can be slowly eliminated after i.t. instillation and do not significantly cross the pulmonary barrier in rats. Six months after a single instillation the MWCNTs were still present, although chemically modified and cleaved.
Surface modification of poly(D,L-lactide-co-glycolide) (PLG) NPs with cationic coatings, especially chitosan (+36.8 ± 0.4 mV), proved to be a sustained gene delivery vehicle in pulmonary epithelial cells compared to weak anionic coatings (−11.0 ± 2.0 and −23.2 ± 1.1 mV); no correlation could be found between the magnitude of zeta potential and NP uptake.60 Xu et al.61 developed a negative-to-positive charge-reversal technique in which NPs (composed of folic acid functionalized polycaprolactone–polyethyleneimine (PCL–PEI)/amide) are negatively charged in a neutral milieu but become positively charged at pH 5–6. Cellular uptake was improved for particles carrying positive charges, thus rendering these particles ideal for tumor and nuclear drug delivery. The study performed by Klesing et al.62 showed that calcium phosphate–PEI core–shell NPs (+32 mV) – unlike the anionic (−26 mV) counterpart – loaded with a photoactive dye efficiently caused, due to cellular uptake, cell death in HIG-82 and J774A.1 cells. Asati et al.63 studied the cytotoxicity of surface-modified CeO NPs in a panel of non-transformed cells (cardiac myocytes, H9c2 and human embryonic kidney cells, HEK293) and transformed cells (lung carcinoma, A549 and breast carcinoma, MCF-7). The NPs without surface modification were not toxic to any of the cell lines, although these particles did enter the cytoplasm in a significant proportion of the cells in the cell culture. The carboxylated (≈−44 mV) NPs were only toxic to the A549 cells (cytoplasmic localization). The positively charged (aminated; ≈ +27 mV) NPs also internalized in all cell types (except MCF-7), but they were preferentially localized in the lysosomes and became toxic due to the activation of oxidase activity of nanoceria by the acidic environment. Liu et al.64 showed that positively charged polystyrene NPs (NH2–PS) displayed a higher cytotoxicity than negatively charged (COOH–PS) particles in HeLa and NIH 3T3 cells. Moreover, the NH2–PS particles caused DNA damage and the activation of checkpoints whereas the COOH–PS had no obvious effect on the cell cycle. Also positively charged silver (+40 mV), gold and silica NPs (charges not mentioned) induced more cytotoxicity in bacterial, Cos-1 and NR8383 macrophage cells, respectively, compared to their negatively charged counterparts (silver: −22, −38 and −10 mV) (Table 1).65–67
Reference | Measuring solution | Cationic coating/formulation | Zeta potential (mV) | Anionic coating/formulation | Zeta potential (mV) | Neutral coating/formulation | Zeta potential (mV) |
---|---|---|---|---|---|---|---|
a NH2: aminated particles; COOH: carboxylated particles; n.a.: not assessed. | |||||||
Baoum et al.60 | KCl solution | PVAm–PLG | +37.5 ± 0.8 | PVA–PLG | −11.0 ± 2.0 | ||
Chitosan–PLG | +36.8 ± 0.4 | Pluronic F-127–PLG | −23.2 ± 1.1 | ||||
DC–chol–PLG | +33.2 ± 0.7 | ||||||
Cetrimide–PLG | +35.4 ± 1.6 | ||||||
Protamine–PLG | +17.7 ± 1.4 | ||||||
DODAB–PLG | +40.9 ± 0.9 | ||||||
Xu et al.61 | n.a. | PCL–PEI/amide | +50 (pH 5) | PCL–PEI/amide | −20 (pH 7.4) | ||
+8 (pH 6) | |||||||
Klesing et al.62 | n.a. | PEI–CaP | +32 | Carboxyl methyl cellulose–CaP | −26 | ||
Asati et al.63 | n.a. | NH2–PNC–CeO | ≈ +27 | PNC–CeO | ≈ −44 | Dextran–CeO | ≈ −1 |
Liu et al.64 | Water | NH2–PS 50 nm | +36.6 | COOH–PS 50 nm | −9.0 | ||
NH2–PS 100 nm | +45.8 | COOH–PS 100 nm | −34.0 | ||||
NH2–PS 500 nm | +37.9 | COOH–PS 500 nm | −38.5 | ||||
Culture medium | NH2–PS 50 nm | +3.7 | COOH–PS 50 nm | −23.6 | |||
NH2–PS 100 nm | +4.2 | COOH–PS 100 nm | −23.7 | ||||
NH2–PS 500 nm | +5.6 | COOH–PS 500 nm | −21.9 | ||||
El Badawy et al.65 | As prepared | BPEI–Ag | +40 | H2–Ag | −22 | ||
Citrate–Ag | −38 | ||||||
PVP–Ag | −10 | ||||||
Goodman et al.67 | n.a. | NH3–MMCP | n.a. | COOH–MMCP | n.a. | ||
Bhattacharjee et al.66 | n.a. | NH2–Si | n.a. | COOH–Si | n.a. | NH3–Si | n.a. |
Schaeublin et al.68 | n.a. | TMAT–Au | n.a. | MES–Au | n.a. | ||
Harush-Frenkel et al.69 | Double distilled water | Stearylamine–PEG–PLA | +31.7 ± 2.1 | PEG–PLA | −30.7 ± 0.9 | ||
Nemmar et al.70 | Phosphate buffer | NH2–PS | −44 | COOH–PS | −41 | ||
Geys et al.71 | Saline | NH2–QDs | −14.2 | COOH–QDs | −35.2 |
However, unlike the aforementioned studies, Schaeublin et al.68 found no correlation between surface charge of gold NPs and cytotoxicity in HaCaT cells. Abnormal cell morphology was observed regardless of surface charge and the positively and negatively charged NPs induced cell death by apoptosis, whereas the neutral gold NPs induced necrosis (charges not reported).
Although different research groups used different NPs and experimental models, it seems that in general, in vitro, positively charged NPs are taken up more easily into the lysosomes and cause more cytotoxicity which is attributed to the acid environment.
Drawing a general conclusion is less evident with regard to the in vivo studies. Choi et al.20 demonstrated in a rat model that NPs with a noncationic surface charge (and smaller than 34 nm) translocate rapidly from the lung to mediastinal lymph nodes. Harush-Frenkel et al.69 administered differently charged copolymer NPs i.t. to mice and observed that the positively charged (+31.67 ± 2.1 mV), but not the negatively charged (−30.7 ± 0.9 mV), NPs increased pulmonary effects, together with a transient systemic toxicity, mainly on white blood cells. In a hamster model, Nemmar et al.70 studied the effects of polystyrene (PS) NPs on thrombus formation after i.v. and i.t. administration. The unmodified NPs had no effect whereas the carboxylated NPs significantly inhibited thrombus formation after i.t. administration. The amine-modified NPs enhanced thrombus formation after i.v. as well as i.t. administration. The observed effects could mainly be explained by platelet activation. Geys et al.71 also observed acute prothrombotic effects after i.v. administration of NPs. In mice, i.v. injection of carboxyl-modified quantum dots (QDs) – but not amine–QDs – caused thrombosis. The number of pulmonary thrombi was proportional to the administered dose. The authors showed that these thrombotic events were caused by activation of the coagulation cascade and not by platelet activation as opposed to that mentioned in the study by Nemmar et al. Although the findings of the last two studies seem to be contradictory, it must be noted that the effect was initiated by dissimilar mechanisms, that the nature of the NPs used was different, and that while the zeta potentials of the carboxyl-modified QD and PS beads (−35.2 mV and −41 mV) were similar, the amine-modified NPs showed very different zeta potentials (−14.2 mV for QDs and −44 mV for PS).
A remarkable finding from these last studies, and others, is that addition of functional groups to the NP surface does not necessarily change the zeta potential to the same direction, e.g. carboxyl-functionalized NPs do not necessarily create a negative zeta potential. Another remark to make here is that most aforementioned studies consider the uptake of their NPs in a certain model system, but provide less information on the toxicity. Due to the functionalization of the NPs used it is difficult to distinguish whether the observed effects are caused by the NP functionalization – with a marked but punctate charge – or overall NP charge.
The zeta potential is one of the many essential NP characteristics to be measured when performing experimental studies. It gives information on the particle's stability in solution. Linking zeta-potential to particle toxicity is more difficult. The zeta potential is an important parameter when determining toxico-kinetics and the bioavailability, but the charge at specific spots (which can differ from the zeta potential value) also contributes significantly to the NP toxicity, making the zeta potential not a predictor of the nanotoxicity.
To understand the biological impact of NPs, a notion of the kinetics of NP–protein association and dissociation is required since these determine the particles' interactions with biological surfaces and receptors. The protein corona can be subdivided into a “soft” component where a dynamic exchange between the NPs and proteins present in the medium exists and a “hard” corona where the adhered proteins have a higher affinity for the particle's surface. Rational reasoning reveals that more abundant proteins probably have a less profound effect than less abundant proteins or proteins with a high specificity for a certain receptor.72,73,75 In an attempt to identify the molecules present in the corona, a set of chemically different NPs was examined and serum proteins such as albumin, fibrinogen and IgG have been found to form the majority of the corona. This is not surprising since albumin and fibrinogen are the most abundant serum proteins and will, therefore, dominate the particle surface for short periods. Other proteins, such as immunoglobulins, complement, apolipoproteins and alpha-1-antitrypsin also have been identified in the protein corona.73,75 A positive consequence of binding by immunoglobulins and components of the complement system to NPs' surface is their opsonization and, hence, enhanced NP phagocytosis, which is the main pathway to remove foreign materials, larger than the renal filtering threshold, from the blood circulation.76 On the other hand, avoiding this pathway might be useful to develop nanomedicines having prolonged circulation times.
A negative consequence of protein adsorption on the NP surface is the potential structural and functional perturbation of these proteins.74 Protein binding might disturb protein self-assembly reactions and protein folding, possibly leading to functional defects (e.g. loss of enzymatic activity), disturbance of biologic processes or precipitation of ordered polymeric assemblies (amyloid fibrils) thus inducing diseases such as amyloidosis.77 For instance, it has been shown in vitro that hydrophilic polymer-coated quantum dots, cerium oxides and MWCNTs can increase the rate of fibrillation of human β2-microglobulin forming multiple layers on the particle's surface, resulting in a local increased protein concentration, precipitation and the formation of oligomers;74 this implies, if confirmed in vivo, the possibility of serious health effects e.g. in the brain.78
Bearing this in mind, a thorough characterization of the protein corona is an important, though difficult to achieve, addition to the “NP characterization list”. It illustrates that not only the pristine particle (as produced) should be characterized, but also NPs that are used in a specific biological assay (under specific experimental conditions) should be looked at in detail.
Several commonly used NPs can appear in different forms having significantly different toxicological profiles. In the following section we will briefly discuss TiO2, different types of carbon-based NPs, and silica. The reason for this selection is that for these NPs a certain number of studies have been performed where different crystal structures are compared within a model system.
The relation between different crystal forms and toxicity is relatively well established in the case of titanium dioxide (TiO2). Rutile TiO2 is the most common natural form, whereas anatase and brookite are rarer forms, although the anatase form can often be found in synthetic TiO2, especially as NP. Other, mostly synthetic, forms also exist like cubic and baddeleyite-like.
Warheit et al.79 compared the pulmonary toxicities of three forms of ultrafine TiO2 NPs in rats, with quartz as a positive control. After i.t. administration of two kinds of ultrafine TiO2 (uf-1 and uf-2), ultrafine 80/20 anatase/rutile (uf-3) or fine rutile TiO2 (F-1), the authors observed pulmonary inflammation, cytotoxicity and fibrogenic effects with quartz and uf-3, whereas the other NPs only produced a transient inflammation. These effects could be related to differences in crystal structure, but other factors such as inherent pH, surface chemical reactivity or production process could not be excluded, moreover when the forms were mixed (uf-3), the observed effects were relatively large. Gurr et al.80 compared the potency between anatase and rutile TiO2 NPs to induce oxidative damage to human bronchial epithelial cells. They reported induction of oxidative DNA damage, lipid peroxidation, micronuclei formation and increased H2O2 and NO production with 10 nm and 20 nm anatase TiO2 treatments in the absence of photocatalysis. Larger sized (>200 nm) anatase TiO2 did not induce oxidative stress and the results with the rutile TiO2 (200 nm) were inconsistent. The anatase–rutile mixture induced, as in the former article, more oxidative DNA damage than the pure anatase and rutile forms. Sayes et al.81 exposed human dermal fibroblasts and A549 cells to similar-sized rutile and anatase TiO2 NPs. The anatase NPs were photocatalytically more potent (producing ROS under UV illumination) than the rutile NPs. These data correlated well with the observed cytotoxicity since the anatase NPs were 100 times more toxic than the rutile NPs. Overall, the anatase TiO2 NPs show more toxicity compared to rutile TiO2 NPs and the mixture of both forms seems to be at least equally toxic as the most toxic component.
Another group of popular NMs consists of those built from carbon comprising engineered as well as naturally occurring particles such as fullerenes (C60 or Bucky Balls), CNTs, carbon black (CB) etc. Most comparative studies have been performed to compare CNT and CB. Teeguarden et al.42 compared the acute pulmonary toxicity of SWCNT, crocidolite asbestos and UFCB after i.t. instillation in C75Bl/6 mice. SWCNTs were the most potent, causing pulmonary inflammation. These SWCNTs also induced the most pronounced fibrotic response. In this study, SWCNT was the most potent compound, more than crocidolite asbestos, and UFCB was the least potent. Lam et al.45 studied the fate of i.t. instilled CNT and CB in B6C3F mice. Both particle types were taken up by alveolar macrophages, but the CB-laden macrophages were scattered in the alveolar space while the CNT-laden macrophages entered alveolar septa and clustered to form epithelioid granulomas. The CB-treated animals did not develop a long-lasting pulmonary inflammation, whereas in the CNT-treated animals this was very pronounced; moreover 90 days after the single exposure the inflammation and necrosis had extended into the alveolar septa. The results of these in vivo studies are very consistent and all show that CNTs are more toxic, compared to UFCB at equal mass doses. In vitro, on the other hand, Garza et al.82 found CB, MWCNTs (and asbestos) to be equally potent at causing cytotoxicity and ROS generation in A549 cells.
There is a growing use of silica-based NMs in commercial products. Crystalline silica (SiO2) exists in multiple forms. Quartz (dense crystalline silica) is a widespread and well-known material that exists in natural and synthetic forms. Porosil is the family name for porous crystalline silica, which exists exclusively in a synthetic form. Zeolites are crystalline aluminosilicate or silicate materials with very large surface areas due to their pores. In contrast to the well-studied crystalline micron-sized silica, very little information exists on the toxicity of its nano-sized forms, as recently reviewed by Napierska et al.83 Wang et al.84 investigated cytotoxicity and genotoxicity of ultrafine crystalline silica (UF-SiO2) in cultured human lymphoblastoid cells and reported a fourfold increase in the frequency of micronucleated binucleated cells compared to the untreated control. However, it should be emphasized that the UF-SiO2 used was extracted from commercially available crystalline silica and the particle sizes were not uniform. Warheit et al.22 compared the in vivo toxicity in rats instilled with synthetic nanoquartz particles (12 and 50 nm), mined Min-U-Sil quartz (500 nm) and synthetic fine-quartz particles (300 nm). Exposure to the quartz particles of different sizes produced pulmonary inflammation and cytotoxicity, with nanoscale quartz of 12 nm and Min-U-Sil quartz being more toxic than fine quartz and nanoscale quartz of 50 nm. The pulmonary effects were not consistent with particle size but were associated with surface activity, particularly hemolytic potential, but the exact nature of the difference was not revealed.
The majority of silica NMs that have found worldwide applications are amorphous. However, recently nano-versions of zeolites have been synthesized, as these nanocrystals present potential benefits over their micron-sized analogues.85 The toxicity of micron-sized zeolites is generally higher than that of micron-sized amorphous silica,86 but less than that of quartz; the internal surface of micron-sized zeolites does not appear to influence their toxicity most likely because their internal surface does not interact with the biological system of interest.85 The cytotoxic activities of synthesized nanozeolites Y and A were recently assessed by Thomassen et al.87 These zeolite NPs induced very low in vitro toxicity compared to the positive control of similar size (60 nm amorphous silica NPs), despite an approximately 10-fold greater specific surface area.
Midander et al.13 studied the toxicity of nano- and micrometer sized copper and copper(II) oxide particles in A549 cells. Next to the size-dependent cytotoxic effect, the authors found a size-dependent release of copper; higher concentrations of copper ions were found in suspensions of nanosized particles. In comparative experiments, it was shown that the cytotoxic effects related to the released copper were significantly lower than the particle related effects. Moreover, the NPs were potent at causing DNA damage, whereas this was not the case for the released copper fraction.
In Balb/c mice, Stebounova et al.89 found only a limited pulmonary inflammatory response after silver NP inhalation (primary size: 5 ± 2 nm). Kim et al.90 observed DNA damage and a cytotoxic response in human hepatic cells induced by oxidative stress, which was prevented in co-incubation with N-acetylcysteine. The authors concluded that the observed effects were caused by the NPs themselves since dissolution assays showed minimal ion dissolution. Several authors have shown that nano-silver exposure leads to silver uptake after inhalation, oral as well as dermal exposure (reviewed by Christensen et al.91), but it is still unclear how the silver – as ions or as NPs – is taken up.
Ponti et al.92 observed that both cobalt NPs and cobalt ions induced cytotoxicity in Balb/3T3 mouse fibroblasts. Genotoxicity and morphological transformations were only induced by the cobalt NPs. Horev-Azaria et al.93 recently studied cytotoxicities of cobalt NPs in A549, MDCK, NCIH441, Caco-2, HepG2 and dendritic cells. The authors observed dissolution of cobalt ions and a cell-specific sensitivity towards cobalt ions and/or cobalt particles. The findings suggest that the observed toxic effects were mainly caused by the dissolution of cobalt, which is in line with the observation of Hoet et al.94 who found that Co-particles freshly immersed in the medium lead to an early stimulation of the hexose monophosphate shunt in rat alveolar type II cells. Lison et al.96,97 showed that the immersion of Co particles results in an immediate production of oxidative radicals. The presented data indicate that the production of hydrogen peroxide rather than the formation of hydroxyl radicals forms the basis of the early onset of oxidative stress. Cobalt oxide, in contrast to Co-metal particles, only releases minimal amounts of Co-ions. Papis et al.98 found a concentration and time-dependent cytotoxicity with cobalt-oxide NPs, mainly caused by the NPs themselves, although the authors could not exclude that Co dissolution occurred inside the cells, since the NPs altered the expression of the same set of genes as did the ions.
Xia et al.95 studied the mechanism of cytotoxicity of zinc oxide NPs in RAW 264.7 and BEAS-2B cells. As expected, ZnO NPs were highly toxic to both cell types inducing oxidative stress, inflammation and cell death. ZnO dissolutions in the culture medium and in the cells were observed, these being assessed by ICP-MS and confocal microscopy using a Zn+ specific dye (Newport green DCF), respectively. Non-dissolved ZnO NPs entered caveolae in BEAS-2B and were found in lysosomes in RAW 264.7 cells. Due to the high toxicity observed with the ZnO NPs, iron-doped ZnO NPs were synthesized and studied in vivo. Doping the ZnO NPs with iron suppressed the toxic effects on zebrafish hatching. In mice and rats, i.t. instillation of the doped ZnO NPs resulted in decreased neutrophil cell counts and IL-6 mRNA production compared to non-doped ZnO NPs. The authors showed that iron doping slows the rate of Zn+ dissolution and might, therefore, be a good strategy to reduce the toxicity of ZnO NPs.99
Iron oxide particles (Fe2O3) of 0.5 and 1.5 μm were studied in vivo and in vitro by Beck-Speier et al.100 The larger, but not the small, particles induced inflammation in the rat lung 24 hours after i.t. instillation. In rat alveolar macrophages also the larger particles only increased the production of IL-6 and caused decrease of PGE-2. The difference in toxicities could be explained by the larger dissolution rate of the smaller particles (0.0037 ± 0.0014 day−1), compared to the larger particles (0.0016 ± 0.0012 day−1). The authors assumed that soluble iron seems to have a protective effect on particle-induced inflammation. Two nanosized iron oxide particles, Fe2O3 (22 nm) and Fe3O4 (43 nm), were studied for their cytotoxic effects in human aortic endothelial cells (HAECs) and monocytes (U937) by Zhu et al.101 Both NPs induced cytoplasmic vacuolation, mitochondrial swelling and cell death in HAECs, together with increased NO production and elevated NOS activity. Monocyte adhesion to the HAECs was significantly enhanced as a consequence of ICAM-1 upregulation and IL-8 production. Although cytotoxicity was observed at high concentrations, HAEC and U937 cell viabilities increased at low Fe3O4 concentrations (2–20 μg ml−1), but not with Fe2O3, which might be explained by the increased dissolution of Fe3O4 in U937 cells, compared to Fe2O3 NPs. Another possible explanation is that Fe3O4 NPs aggregate and deposit on HAECs thereby decreasing bioavailability. So, at low concentrations (below 150 ng ml−1), soluble iron can have a protective (stimulatory) effect but at high concentrations induce cytotoxicity.
In several of the aforementioned studies, the authors linked the presence of the metal ions with observed toxicity but it should be taken into account that also the process of dissolution can drive the toxicity. Metal ion dissolution creates electrical instabilities in the particle's matrix (redox reactions) thus increasing the particle's reactivity and probably also its toxicity.
In conclusion, iron oxide particles – and low concentrations of dissolved iron, due to the low dissolution rate – do not cause significant cytotoxicity, whereas silver ions cause in vitro cytotoxicity, but seem quite harmless in vivo. The dissolution of cobalt metal is a process that induces clear oxidative stress. Zinc ions, arising from ZnO particles, are cytotoxic in vitro – and less in vivo – which can probably be explained by the balanced homeostasis of these ions in whole organisms compared to cells in culture.
In the search for the most important nano-characteristic(s) in toxicology numerous investigations have been undertaken during the last two decades. The quest has, however, not been too successful. Products containing NPs were already on the market for several years before reports on biological adverse effects started to appear. So, unfortunately our knowledge on the possible hazardous effects of nanotechnological applications limps behind in the technological progress. Initially, researchers in health sciences underestimated the complexity of NMs and experiments were set up in a similar way as with chemicals. Later it became clear that both the nature of the material examined and the assays used should be reviewed in detail before meaningful data could be produced. Now, most researchers are aware of the difficulties, and the scientific community has started to fill the data gaps. Therefore, it is to date very difficult – if not impossible – to link NP characteristics to their toxicity. This is partly caused by the enormous variety of NPs that has been tested without (in the early years) appropriate material characterization and standardized dispersion and experimental protocols. This makes it very difficult to compare results of different experimental studies and has hampered the identification of clear relations between certain NP characteristics and their biological effects.
The interplay between the different NP properties and also interactions with their environment determine the biologic activity and toxicity of NPs. In Fig. 3 the most important characteristics are depicted. As mentioned above, size and aggregation/agglomeration are well related; during NP dispersion individual NPs might aggregate/agglomerate due to proteins and/or ions present in the mixture. Dissolution is also related to NP size (shape and crystal structure). In general, smaller NPs have higher dissolution rates compared to larger particles of the same composition. Depending on the shape and crystal structure (irregularities), the free energy of the particle is distributed differently, affecting e.g. dissolution and redox-reactions. NP charge, protein corona and reactivity are also closely related. Depending on the surface charge (and functionalities) different kinds of proteins present in the dispersant, culture medium or biological fluid will adhere to the NP surface thereby blocking particular parts of the NP surface. As a consequence, the particle will interact differently with its surrounding milieu and maybe with this change its reactivity and aggregation properties. Clearly, all these parameters contribute to the biological activity and toxicity of engineered NPs, but the exact mechanisms and how the interplay of these parameters determine toxicity remain far from elucidation. This points out the need for validated testing strategies so that hazardous NPs can be identified and subsequently categorized according to their potential toxicities. Ultimately, an algorithm could be developed which identifies hazardous NPs according to their physico-chemical properties.105
![]() | ||
Fig. 3 Important physico-chemical properties for NP characterization and their interrelations. |
In addition to the routine characterization such as morphology, chemical composition, shape, specific surface area, primary particle size and size distribution it was mentioned by Fubini et al. that phenomena occurring during the contact between NPs and cellular media or biological fluids (dispersion, agglomeration/aggregation, protein adsorption) need consideration.106 Therefore it is important to perform independent NP characterization since different dispersion protocols lead to different NP characteristics and hence the information provided by the manufacturer might differ from the researchers'. But sometimes the information provided by the manufacturer might be erroneous as described in several papers e.g. Park and Grassian107 and Park et al.108 found a significant difference between their independent characterization data and the manufacturers' information on the NPs used in their experiments. Possible explanations were a lack of standards for sample measurements, methods and protocols, and batch-to-batch variability.
In view of the fact that the number of products containing engineered NPs increased every year during the last decade it is expected that the general population will be exposed to the NPs. The most obvious exposed population are workers, mainly during manufacturing and bagging and un-bagging of products. Some reports describing adverse effects of occupational exposure to NMs have already been published, as reports of occupational exposures to mixtures containing nanoparticles in factories in China and India,109,110 and case-reports regarding a worker who inhaled an estimated gram of nickel nanoparticles111 and a chemist exposed to high levels of intermediate or final products of dendrimers while performing dendrimer synthesis.112
After sale, all users are potentially exposed. As many NPs are incorporated in a matrix, exposure will occur during the wear or breakage of the products. Consequently, it is difficult to estimate and/or measure the dose to which different populations might be exposed. To date, it is not only difficult to estimate the doses we are exposed to, but also no NP exposure–disease relationships have been defined and hence no occupational exposure limits (OELs) for NPs–NMs could/have been established (OELs for the bulk form of a material are not appropriate for the NP form, which is obvious due to the above-mentioned reasons).113 Until such limits are defined it is important to protect workers from NP exposure. The first step in a NP risk management program is to identify the exposure scenarios. Afterwards, the exposure concentrations can be detected and control measures can be applied in order to limit the emissions during these processes/tasks.
The National Institute for Occupational Safety and Health (NIOSH) conducted on-site evaluations to examine potential airborne release of NMs in facilities where engineered NMs are produced and/or handled, using the Nanoparticle Emission Assessment Technique (NEAT).114,115 First, the potential sources of NM emissions were identified using the condensation particle counter (CPC) and optical particle counter (OPC), both direct-reading, portable instruments to detect NP emissions (measure particle number concentrations). Background measurements were subtracted from the measurements made during the examined process/task. Afterwards filter-based air sampling was performed to identify the particles and subsequently chemical and microscopic analysis was done for chemical speciation. The results from 12 field studies (research and development laboratories synthesizing/handling CNTs, fullerenes, metal oxides and QDs, pilot plants producing carbon nanopearls and metal oxides and facilities that incorporate engineered NMs into a final product or create NMs as part of a process) showed that NM emissions varied among the different facilities and that they could be detected using the NEAT strategy. This strategy also allowed evaluation of the effectiveness of the control measures – e.g. ventilation, confinement – taken.
At the moment we lack clarity as to which characteristics are important for determining potential hazards. The existing exposure and/or monitoring techniques – such as electron microscopy, dynamic light scattering and field flow fractionation – are often very sophisticated and lack direct data output and portability and are therefore not suited for routine exposure monitoring.116
Now, several research groups work on an inexpensive “universal aerosol monitor” suitable for personal use which can simultaneously measure particle numbers, mass and surface area. Such strategies can e.g. be found in the following research papers.117–121
Taking all together, this overview calls for future studies to use tailored NPs having the same chemical composition, differing in only one (or a limited number of) physico-chemical characteristic, so that the results obtained give more clear-cut information; and also highlight the importance of a thorough NP characterization.
AR | Aspect ratio |
BPEI | Branched polyethyleneimine |
CB | Carbon black |
CHO | Chinese hamster ovary |
CNT | Carbon nanotube |
CPC | Condensation particle counter |
CTAB | Cetyl trimethylammonium bromide |
ESR | Electron spin resonance |
HAEC | Human aortic endothelial cells |
ICAM-1 | Intercellular adhesion molecule-1 |
ICP-MS | Inductively coupled plasma-mass spectrometry |
IL | Interleukin |
i.t. | Intratracheal |
i.v. | Intravenous |
LPS | Lipopolysaccharide |
MES | 2-Mercaptoethanesulfonate |
MMCP | Monolayer protected gold clusters |
MWCNT | Multi-walled carbon nanotube |
NEAT | Nanoparticle emission assessment technique |
NIOSH | National institute for occupational safety and health |
NOS | Nitric oxide synthase |
NM | Nanomaterial |
NP | Nanoparticle |
OEL | Occupational exposure limit |
OPC | Optical particle counter |
PCL | Polycaprolactone |
PEI | Polyethyleneimine |
PEG | Polyethyleneglycol |
PGE-2 | Prostaglandin E−2 |
PLG | Poly(D,L-lactide-co-glycolide) |
PNC | Poly(acrylic acid) |
PS | Polystyrene |
PVP | Polyvinylpyrrolidone |
QD | Quantum dot |
RBC | Red blood cell(s) |
ROS | Reactive oxygen species |
SWCNT | Single walled carbon nanotube |
UF | Ultrafine |
This journal is © The Royal Society of Chemistry 2013 |