Nanomaterials

5. What are the potential health effects of nanomaterials?

• 5.1 How do nanoparticles interact with proteins?
• 5.2 How can nanomaterials be transported in the body?
• 5.3 What are the potential effects of carbon nanotubes?
• 5.4 Can nanomaterials have genetic effects?
• 5.5 Do nanoparticles have effects on the heart and blood vessels?

Many of the currently available OECD guidelines for the testing of chemicals are likely to be adequate to detect potential hazards of manufactured nanomaterials as well (SCENIHR 2007a). However, considering the particulate nature of the manufactured nanomaterials some adaptation of the testing methodology is likely needed (SCENIHR 2007a). This is especially the case for the expression of the dose metric as administered in the test systems. The currently available information in the use of the OECD guidelines for the evaluation of manufactured nanomaterials is, however, limited. Warheit et al. (2007) reported on a base set of toxicity tests for detection of the acute toxicity of ultrafine TiO2 particles using assays as described in various OECD guidelines. The results of most of the studies demonstrated low hazard potential in mammals or aquatic species following acute exposures to the ultrafine TiO2 particle-types tested. In the studies particle sizes were approximately 140 nm in diameter when TiO2 was dispersed in water, but increased up to approximately 2000 nm when present in phosphate buffered saline (PBS), thus indicating the importance of nanoparticle characterisation as they are used in various test conditions (see section 4.2). Recently, the OECD has started a sponsorship programme in which, for 14 of the most used nanomaterials, a dossier on hazard identification will be produced (OECD 2008a). In this programme the applicability of the various OECD guidelines for nanomaterial testing will also be evaluated.

In the possible applications of validated in vitro assays for determination of general toxic effects of nanomaterials there has been almost no progress in relation to risk assessment. There is still a clear need for validated in vitro assays for nanoparticle evaluation. In the base set as reported by Warheit et al. (2007), only two in vitro assays were used, both of which are assays for the detection of genotoxicity. One assay (Ames test) uses bacterial cells and the other (chromosomal aberration test) uses mammalian cells. For the bacterial assays there can be reasonable doubt whether the manufactured nanomaterials in the size range used (140 nm) can enter the bacterial cells. Recent results reported by Sayes et al. (2007) did not support the use of in vitro assays for toxicity endpoints. For five different particle types the range of toxicity end points showed little correlation between in vitro and in vivo measurements for inhalation toxicity profiles (Sayes et al. 2007). Recent evaluations of safety assessment of manufactured nanomaterials indicated that in vitro assays may be useful, but mainly for screening and the evaluation of specific mechanistic pathways (ECETOC 2006, Oberdörster et al. 2005a). So, they may be used for assessing the possible reactivity, inflammatory potential and celluar uptake of nanoparticles. However, to be applicable in risk assessment, these assays need to be validated and their relevance for in vivo hazard identification needs to be demonstrated.

It should be noted that the in vivo assays as described in the various OECD guidelines are not validated for nanomaterials either. However, the experience gained in the testing of chemicals with these assays indicates that they can be used for the detection of some potential human and ecological hazards.

5.1 How do nanoparticles interact with proteins?

As the protein corona may affect the nanoparticle behaviour including its biological effect, the nanoparticle may also have an effect on protein behaviour. Nanoparticles were found to have the potential to promote protein assembly into amyloid fibrils in vitro by assisting the nucleation process (Linse et al. 2007). This phenomenon may have implications for human disease as protein self-assembly of a variety of proteins and peptides is known to cause human amyloid disease (Chien et al. 2004, Chiti and Dobson 2006, Koo et al. 1999). Large insoluble protein fibrils are formed resulting in amyloid plaques, an example being dialysis related amyloidosis due to β2-microglobulin (Floege and Ehlerding 1996).

Various types of nanoparticles (copolymer nanoparticles, cerium oxide particles, quantum dots, and carbon nanotubes) were found to enhance the probability of appearance of a critical nucleus for nucleation of protein fibrils from human β2-microglobulin (Linse et al. 2007). The shorter nucleation phase depended on the amount and nature of the particle surface. There was an exchange in protein on the particle surface in which β2- microglobulin formed multiple layers on the particle surface providing a locally increased protein concentration promoting oligomer formation. These results suggest a mechanism involving surface assisted nucleation that may increase the risk for toxic cluster and amyloid formation (Linse et al. 2007). Further research demonstrated that besides an increase in nucleation, nanoparticles may also retard the nucleation process which was attributed to an effect on the nucleation step of the amyloid beta protein while the elongation step was unaffected (Cabaleiro-Lago et al. 2008). These experiments were performed using an incubation of nanoparticles with purified β2-microglobulin protein. Recently this effect of a shortening of the nucleation process was also demonstrated for β- amyloid after incubation with TiO2 nanoparticles (Wu et al. 2008). Whether the observed nucleation process also occurs in an in vivo situation or in more complex biological fluids where competitive binding may take place remains to be determined.

There are indications that after deposition at the olfactory mucosa of the nose nanoparticles may translocate into the brain (see below). This observation raises some concern in view of the amyloid diseases of the brain in the context of the potential of nanoparticles to cause protein fibrillation in vitro. This is certainly an area for which additional research is urgently needed.

5.2 How are nanomaterials detected and analysed?

Toxicokinetics is the science dealing with absorption, distribution, metabolism and excretion (ADME) of substances in the body. This whole cascade of events occurring after an (external) exposure determines the internal exposure of organs at risk to potential toxic substances. Prominent exposure routes of nanoparticles are inhalation, ingestion and skin uptake as well as intravenous injection for medical purposes.

Translocation of manufactured nanoparticles through the epithelium is likely to depend on the physical-chemical properties of the nanoparticle, e.g. surface charge, hydrophobicity, size, presence or absence of a ligand, and physiology of the organ of intake e.g. healthy vs diseased state (where translocation may be increased or decreased depending on the illness) (Des Rieux et al. 2006). Under normal physiological conditions, paracellular transport of nanoparticles would be extremely limited, as pore size at tight junctions is between 0.3-1.0 nm (Des Rieux et al 2006). Little is known on the behaviour and fate of nanoparticles in the gastrointestinal tract (EFSA 2008).

While some mechanisms may be of general applicability for many biological membranes, it must be noted that each membrane has specific tasks so that mechanisms related to those tasks may not be applicable to other membranes. After translocation/absorption, the distribution of nanoparticles inside the body over the various organ systems and within the organs needs to be determined. After the initial translocation/absorption of nanoparticles the systemic circulation can distribute the particles to all organs and tissues in the body.

As a model particle for nanotechnology research including toxicokinetic studies, metallic colloidal gold nanoparticles are widely used. They can be synthesised in different forms (rods, dots), are commercially available in various size ranges, and can be detected at low concentrations. Human cells can take up gold nanoparticles without cytotoxic effects (Connor et al. 2005). In particular for biomedical applications, they can be considered relevant models, since they are used as potential carriers for drug delivery, imaging molecules and even genes (Kawano et al. 2006), and for the development of novel cancer therapy products (Hainfield et al. 2004, Hirsch et al. 2003, Loo et al. 2004, O’Neal et al. 2004, Radt et al. 2004). In addition, gold nanoparticles have a history as labels for tracking protein distribution in vivo in which proteins are coupled to small colloidal gold beads at nanoscale dimensions (Heckel et al. 2004, Hillyer and Albrecht 1999).

For systemic distribution, direct systemic exposure of organs can be obtained by the intravenous route for which the total internal dose/exposure is equal to the administered dose. Distribution of particles occurs at multiple organs including liver, spleen, heart and brain (De Jong et al. 2008, Ji et al. 2006).

When rats were intravenously injected with solutions containing various sizes of metallic colloidal gold nanoparticles (10, 50, 100 and 250 nm), the distribution of gold nanoparticles was found to be size-dependent, the smallest particles showing the most widespread organ distribution including blood, heart, lungs, liver, spleen, kidney, thymus, brain, and testis (De Jong et al. 2008). The larger nanoparticles mainly resided in spleen and liver. Intravenously injected gold nanorods (length 65 ± 5 nm; width 11 ± 1 nm) accumulated within 30 min, predominantly in the liver. The PEGylation (coating with polyethylene glycol) of these gold nanorods resulted in a prolonged circulation (Niidome et al. 2006).

When negatively charged 1.4 nm or 18 nm gold nanoparticles were injected intravenously, the 18 nm nanoparticles showed a similar pattern as described above with the highest accumulation in liver and spleen. For the 1.4 nm gold nanoparticles only half of the injected dose was present in liver and spleen, whereas the other half was found at higher fractions in the other organs mentioned above, in soft tissues and in the skeleton. Furthermore, the 1.4 nm gold nanoparticles were still circulating in the blood after 24 hours (Semmler-Behnke et al. 2008). When the same gold nanoparticles were injected intravenously in pregnant rats in their third trimester, both 1.4 nm and 18 nm particles were found in the placenta and the foetuses (Semmler-Behnke et al. 2007). For 5 and 30 nm gold colloid solutions very small fractions were found to be transferred to the rat fetus after intravenous administration (Takahashi and Matsuoka 1981). In contrast, transfer to fetal tissue could not be demonstrated by Sadauskas et al. (2007) when gold nanoparticles of 2 nm and 40 nm were injected in pregnant mice.

For intravenously administered TiO2 nanoparticles in rats with a dose of 5 mg/kg body weight and a size range 20-30 nm, the tissue content of TiO2 was determined 1, 14, and 28 days after administration (Fabian et al. 2008). There were no detectable levels of TiO2 in blood cells, plasma, brain, or lymph nodes. The TiO2 levels (μg/g organ) were highest in the liver, followed in decreasing order by the levels in the spleen, lung, and very low in the kidney, and highest on day 1 in all organs. TiO2 levels were retained in the liver for 28 days; there was a slight decrease in TiO2 levels from day 1 to days 14 and 28 in the spleen, and a return to control levels by day 14 in the lung and kidney. A limitation of this study is that most of the particles administered were in the fine fraction up to 1 μm, whereas only 10% by weight was in the nanosize range (<100nm). The results obtained with studies using intravenous administration show that the main target organs for particles are the spleen and liver which are abundant in phagocytic cells i.e. macrophages and Kupffer cells. As was demonstrated for 10 nm particles, the smallest nanoparticles may also show distribution to other organs besides the liver and spleen.

Oral administration of metallic colloidal gold nanoparticles of decreasing size (58, 28, 10 and 4 nm) to mice resulted in an increased distribution to other organs indicating a higher uptake with diminishing size (Hillyer and Albrecht 2001). The smallest particle (4 nm) administered orally resulted in an increased presence of gold particles in kidney, liver, spleen, lungs and even the brain. The biggest particle (58 nm) tested was detected almost solely inside the gastrointestinal tract. For 13 nm sized gold colloids, the highest amount of gold was observed in liver and spleen after intraperitoneal administration (Hillyer and Albrecht 1998). One might speculate whether such translocation of nanoparticles is accompanied by transport of food components/molecules and thus may create an (unwanted) port of entry, which may result in unexpected toxicity or other adverse affects such as induction of allergy.

TiO2 particles of 500 nm were observed in all major tissues of the Gut Associated Lymphoid Tissue (GALT) including Peyer’s Patches and mesenteric lymph nodes after repeated oral administration by gavage for 10 days and evaluation at day 11 (Jani et al. 1994). Systemic exposure occurred as titanium and was detected by chemical analysis in blood, liver, lungs, spleen and heart. The presence of the TiO2 particles was confirmed by histology in Peyer’s Patches, mesenteric lymph nodes, liver, spleen and lung. In the heart the presence of particles could not be confirmed by histology. The highest levels Ti (μg per g tissue) were present in the lymphoid tissues like Peyer’s Patches, mesenteric lymph nodes and the mesentery network. The colon with lymphoid tissue as the appendix and diffuse lymphoid aggregates showed a high Ti level as well. It was concluded that the 500 nm TiO2 particle uptake was primarily taking place via the Peyer’s Patches (Jani et al. 1994).

In a 28 day oral toxicity study of silver nanoparticles (average 60 nm) a dose dependent accumulation of silver was observed in all organs examined, i.e. blood, brain, kidneys, liver, lungs, stomach and testes (Kim et al. 2008). The highest levels were observed in the stomach, followed by kidney and liver, lungs, testes, brain and blood. In the mid and high dose treated groups all levels measured were significantly increased compared with the non treated control group. The silver content was determined by atomic absorption spectrophotometry (AAS) while histology for the confirmation of the presence of silver particles was not performed. Silver levels in the kidneys were, for all doses investigated, twice as high in female rats than in male rats which could not be explained by the results presented (Kim et al. 2008).

It can be concluded that for some nanoparticles the size may be a limiting factor in the potential to cross the GI tract barrier, while for other nanoparticles a similar size may result in uptake from the GI tract which may occur at sizes up to 500 nm.

Several inhalation studies in rodents have shown the distribution of nanoparticles to numerous organs including the liver, the spleen, the heart and the brain (Kreyling et al. 2002, Oberdörster et al. 2002,, Semmler et al. 2004; Semmler-Behnke et al. 2007).

The translocation of radiolabeled insoluble iridium nanoparticles was monitored after inhalation (Kreyling et al. 2002). For both the 15 nm and for the 80 nm particles, most of the particles remained in the lungs from which they were predominantly cleared via the airways into the GI tract and the feces. After systemic uptake from the lungs minimal particle translocation of <1% was observed to secondary organs such as the liver, spleen, heart and brain. The translocation of the 80 nm particles was about an order of magnitude lower than that of the 15 nm particles (Kreyling et al. 2002). Two studies (Semmler et al. 2004, Semmler-Behnke et al. 2007) report on long-term nanoparticle biokinetics in secondary target organs over six months after a single short-term nanoparticle inhalation. Only about 1% of the inhaled nanoparticles had crossed the air- blood-barrier and accumulated in secondary target organs (liver, spleen, kidneys, heart and brain) as well as in the soft tissue and bone. After a transient maximum in all secondary target organs between 1-2 weeks after inhalation, nanoparticle concentrations remained surprisingly constant between week 3 and six months. Unexpected back-and- forth trafficking was observed across the alveolar epithelium of the lungs of the 20 nm sized iridium nanoparticles in the lungs of adult healthy rats after a single one-hour inhalation (Semmler-Behnke et al. 2007). Although being retained in interstitial spaces Ir-nanoparticle re-appeared on the epithelium during the next six months to be cleared by macrophage-mediated transport towards the mucociliary escalator and to GI-tract after swallowing.

When negatively charged 1.4 nm or 18 nm gold nanoparticles were intratracheally instilled into rats, the distribution determined 24 hours later, showed a much higher fraction (8% of instilled dose) of the 1.4 nm particles to be translocated into the circulation and accumulated in secondary target organs than a 30-fold lower fraction of 18 nm particles (Semmler-Behnke et al. 2008). Particles were found in the liver, the spleen, the kidneys, the heart, the brain, the reproductive organs, soft tissue and the skeleton, but by far the highest fraction remained in the lungs. Furthermore, in one other study for gold nanoparticles with sizes of 5-8 nm the majority of the inhaled nanoparticles remained in the lung, although there was a very small but significant fraction being translocated to the blood (Takenaka et al. 2006). When rats were exposed by inhalation for 5 days to gold nanoparticles (majority with a size <35 nm) gold was only detected in the lung and olfactory bulb (Yu et al. 2007). After 15 days of exposure gold could also be detected in other organs including heart, liver, pancreas, spleen, kidney and testis (Yu et al. 2007).

In mice exposed by inhalation to fluorescent Fe containing magnetic nanoparticles (size 50 nm) for four weeks (4h/day, 5 days/week) the nanoparticles were distributed to various organs including liver, spleen, lung, testis and brain (Kwon et al. 2008). The results indicated that both the blood brain barrier(BBB) and the blood testis barrier (BTB) were penetrated by the nanoparticles.

Rats were exposed to carbon nanoparticles (size 20-29 nm) labelled with stable isotope C13 in a whole body inhalation chamber (Oberdorster et al. 2002). Only at high exposures did the label start to accumulate in the liver 30 minutes after exposure. 18 and 24 h after exposure the liver contained about five times more label than the lung. No significant increase in label was observed in the other organs examined which included heart, olfactory bulb, brain and kidney. It is worth noting that 13C is naturally present in each organism at a level of about 1%. Therefore, the deposited 13C labeled nanoparticles represented less than the endogenous 13C content in the mouse lungs.

Ambient air particles and nanoparticles have also been demonstrated to translocate to the brain after inhalation, and thus may potentially influence the central nervous system Using 13C as model particles with a diameter of 36 nm translocation into the olfactory bulb was indicated most likely originating from entry in the olfactory mucosa in the nose (Oberdörster et al. 2004). Also for gold nanoparticles translocation to the brain was observed after inhalation exposure (Yu et al. 2007). When 15-20 nm iridium nanoparticles were administered by inhalation up to 6 months after exposure iridium could be detected in the brain (Kreyling et al. 2002, Semmler et al. 2004). Whether this was solely due to translocation via the nasal absorption is not certain as systemic distribution was also observed in the study. For diesel exhaust containing a nanosized fraction, inhalation exposure resulted in changes in brain activity as demonstrated by changes in EEG signals (Crüts et al. 2008). These observations on possible direct translocation of inhaled nanoparticles in the brain warrant further research to either confirm or reject the hypothesis of nanoparticle association with various brain diseases.

In humans, most inhaled carbon nanoparticles remain in the lung (Brown et al. 2002, Mills et al. 2006, Möller et al. 2008, Wiebert et al. 2006a, Wiebert et al. 2006b). Translocation was found to be <1%. So, although the translocation of nanoparticles from the lungs may occur after inhalation, most nanoparticles remain in the lung and only a minute fraction may reach the circulation. Other efforts to study this translocation from the lung across the air-blood-barrier in humans failed because the experimental limits of detection were about 1% of the administered dose to the lungs and hence above the translocated fraction if at all existing (Brown et al. 2002, Mills et al. 2006, Möller et al. 2008, Wiebert et al. 2006a, Wiebert et al. 2006b). Studies using radiolabelled nanoparticles may have their limitations as it is known that a dissociation of the label from the nanoparticles may occur as shown for the radio-tracer 99mTc when used for labelling carbon nanoparticles (Nemmar et al. 2002).

The results obtained with inhalation studies show that there is the potential of the smaller nanoparticles to cross the air-blood barrier and enter the systemic circulation. In general only a very small portion of the inhaled dose (< 1%) shows translocation. It should be noted that after inhalation exposure, systemic distribution to other organs may also be due to secondary gut uptake after removal of particulates from the lung by the mucociliary mechanism. However, with repeated exposures and the low particle fraction showing migration, it could mean that based on particle numbers a considerable internal systemic exposure may occur.

A prominent clearance pathway of ingested nanoparticles is fecal excretion since every nanoparticle which is not absorbed by the gut epithelium will leave the body via this pathway. Similarly inhaled nanoparticles which deposited on the airways of the respiratory tract will be transported by mucociliary action to the larynx from where it will be swallowed entering the GI-tract. Furthermore, even insoluble nanoparticles deposited in the lung periphery (alveoli) may eventually be cleared by macrophage mediated transport to the distal end of the mucociliary escalator from where they are cleared as described above. However, note that in the human body only 20-30% of the peripherally deposited nanoparticles leave the lungs via this route (Kreyling and Scheuch 2000).

If nanoparticles enter the systemic circulation there are two potential clearance pathways for excretion:

1 Glomerular filtration in the kidneys towards the bladder into urine. In fact, Choi et al. (2007) have observed that intravenously injected quantum dots with a size below 4.5 nm and not bound to any protein due to cystein surface modification will be quantitatively excreted in urine using a rat model. Furthermore, the same clearance pathway into urine applied to intravenously injected, hydrophilically functionalised and positively surface charged SWCNT and MWCNT of up to 2 μm in length was observed in a guinea pig model (Singh et al. 2006). Besides these studies 0.09 and 0.001% fractions of intravenously injected gold nanoparticles of 1.4 and 18 nm size, respectively, were excreted in urine (Semmler-Behnke et al. 2008)

2. Another potential pathway might be a hepato-biliary clearance of nanoparticles from the liver via the bile into the intestine and feces. While this clearance pathway is well known in pharmacology it is only postulated for nanoparticles.

Existing data show that nanoparticles can enter circulation from the respiratory tract or the gastro-intestinal tract. These processes are likely to depend on the physical-chemical properties of the nanoparticles such as size and on the physiological state of the organs of entry. The translocation fractions seem to be rather low; however, this is subject of current intense research.

After the nanoparticles have reached the blood circulation, the liver and the spleen are the two major organs for distribution. Circulation time increases drastically when the nanoparticles are hydrophilic and their surface is positively charged. For certain nanoparticles all organs may be at risk as, for all organs investigated so far, either the chemical component of the nanoparticles or the nanoparticles themselves could be detected, indicating nanoparticle distribution to these organs. These organs include the brain and testis/the reproductive system. Distribution to the foetus in utero has also been observed. As the knowledge of the long-term behaviour of nanoparticles is very limited, a conservative estimate must assume that insoluble nanoparticles may accumulate in secondary target organs during chronic exposure with consequences not yet studied. There is a specific concern considering the possible migration of nanoparticles into the brain and unborn fetus. Research in both of these areas has to be conducted in order to either confirm or reject the hypothesis of nanoparticle association with various brain diseases, and the possible reprotoxic effects of nanoparticles.

5.3 What are the potential effects of carbon nanotubes?

The superficial resemblance between carbon nanotubes and some other high aspect ratio (long thin) nanoparticles was commented upon early in concerns over the safety of nanotubes (Donaldson et al. 2006, The Royal Society and The Royal Academy of Engineering 2004). The term HARN, or High Aspect Ratio Nanoparticles, has been used to cover such structures. For fibrous-type or asbestos-like effects the major concern is mesothelioma, an unusual endpoint that is difficult to study but arises with unusual specificity to certain fibre exposure. Because of the low rate of mesothelioma following inhalation exposure in rats (a few % similar to the prevalence in exposed humans) direct exposure of the peritoneal mesothelium by intraperitonal injection was developed as an assay in the eigthties (Miller et al. 1999, Pott 1995). Takagi et al. (2008) used this approach to examine the tumerogenicity of carbon nanotubes in p53-deficient mice.

Although Takagi et al. (2008) showed that both asbestos and carbon nanotubes caused mesotheliomas whilst fullerenes did not the methodology was criticised (Donaldson et al. 2008, Ichihara et al. 2008) on several gounds including i) the presentation of the test sample indicating that clumps of nanotubes were used that were hundreds of microns in diameter, ii) the dose of 3 mg that was injected into each mouse to obtain the dose of 109 fibres which was originally developed for rats but not mice and iii) the use of the p53 deficient mouse model for mesothelioma detection without using relevant controls.

Poland et al. (2008) used the mouse peritoneal model of direct mesothelial exposure. They tested the acute inflammogenic effects of carbon nanotubes and their ability to cause granulomas on the surface of the diaphragm after one week. The acute response in the peritoneal cavity, inflammation and granuloma formation, mimics the long-term mesothelioma development (Davis et al. 1986, Kane 2006). Various controls were used including long and short amosite asbestos samples that had been used in the 1980's and which produced mesotheliomas in the peritoneal cavity in the case of the long and none in the case of the short fibers (Davis et al. 1986) Nanoparticullate carbon black, graphene in compact form, as opposed to graphene in a tubular form as in carbon nanotubes, was also used. Two carbon nanotube test samples containing a fraction of long straight nanotubes were compared with two samples comprised of short or tangled nanotubes. Inflammation and granulomas were only found in the case of the long straight nanotubes whilst the short/tangled nanotubes had no effect. This property of length-dependent inflammation and early fibrosis in a mesothelial exposure model was therefore shared by both asbestos and the specific carbon nanotubes investigated.

This provides a first support for the contention that specific types of long carbon nanotubes may be pathogenic, like hazardous asbestos, when they have similar characteristics, such as length, rigidity and biopersistence. Whether this poses a risk for humans would depend on whether there is inhalation exposure to these specific types of carbon nanotubes. In addition, the risk would also depend on the possibility for natural migration of nanotubes to the pleural mesothelium from the airspaces. It needs sufficient long straight CNTs to get airborne in workplaces to reach a threshold dose followed by translocation to the pleural mesothelium.

Pacurari et al. (2008) used mesothelial cells in culture and compared carbon nanotubes to asbestos showing that carbon nanotubes induced activation of molecular signaling pathways associated with oxidative stress, similarly to asbestos. The above studies focus on the fibre paradigm for predicting carbon nanotubes effects. Of course carbon nanotubes occur (probably predominantly) as tangled ‘particles‘ of nanotubes material and not as ‘fibres’. So like a ball of string that can be very long yet fit in the hand, carbon nanotubes can be long but be compactly tangled into particles. There is reason to think that such nanotubes pose a ‘particle-type’ hazard that is greater than would be anticipated and reviewed in Donaldson et al. (2006). Extensive research on the effects of purified and non-purified single walled carbon nanotubes on the respiratory tract of mice has been carried out at NIOSH, Pittsburg, USA. The exposure of C57BL/6 mice to non-purified single walled carbon nanotubes (iron content of 17.7% by wt) at 5 mg/m3, 5 hr/day for 4 days was compared with pharyngeal aspiration of varying doses (5-20 μg/mouse) of the same but purified single walled carbon nanotubes. Both exposure regimens resulted in the development of multifocal granulomatous pneumonia and interstitial fibrosis. Non-purified single walled carbon nanotubes inhalation was more effective than aspiration of purified single walled carbon nanotubes in causing inflammatory response, oxidative stress, collagen deposition and fibrosis as well as mutations of K-ras gene locus in the lung of C57BL/6 mice (Shvedova et al. 2005, Shvedova et al. 2008a). Sequential exposure to the same single walled carbon nanotubes and bacteria enhanced pulmonary inflammation and infectivity (Shvedova et al. 2008b). In in vitro studies on a RAW 264.7 macrophage cell line the same purified and non-purified single walled carbon nanotubes showed that the presence of iron in single walled carbon nanotubes may be important in determining redox- dependent responses of macrophages (Kagan et al. 2006)

When nanotubes, possibly of any chemical composition, have similar characteristics as some types of hazardous asbestos, it was demonstrated that similar inflammatory reactions can be induced by the nanotubes as asbestos. The main characteristics required for this to occur are long thin fibrous forms (length >20 micrometer), rigidity, and non-degradability (biopersistence). Whether such nanotubes would pose a risk for humans is unknown, as besides these specific nanomaterial characteristics, inhalation exposure to such structures would also be essential. In addition, migration of such fibrous nanomaterials from the airspaces in the lung to the pleural mesothelium has to occur. In terms of occupational safety, the local air concentration also needs to be higher than threshold doses. The main conclusion of the studies on carbon nanotubes relating to a risk for mesothelioma is that such a risk cannot be excluded. So, when manufacturing nanotubes (possibly of any chemical composition) one should be aware that certain characteristics may pose such a risk and thus should be considered in the safety evaluation of that particular manufactured nanomaterial. Carbon nanotubes seem to conform to the same paradigm as some forms of asbestos, glass fibres etc., that any long, thin biopersistent fibre poses a potential mesothelioma hazard. This means that other high aspect ratio nanoparticles such as nanowires or nanorods are likely to have the same hazard if they satisfy the criteria of length and biopersistence.

5.4 Can nanomaterials have genetic effects?

The genotoxic effects of conventional particles are driven by two mechanisms – direct genotoxicity and indirect (inflammatory processes-mediated) genotoxicity, as reviewed by Schins et al. (2007). Nanoparticles may act via either of these pathways since they may cause inflammation (see above) and they can also enter cells and cause oxidative stress (Donaldson et al, 2005, Nel et al. 2006, Oberdörster et al. 2005a, Oberdörster et al. 2005b, Stone et al. 2007). There is some evidence that the small size may allow nanoparticles to penetrate into sub-cellular compartments that normally exclude environmental particles, like the mitochondria, and nucleus (Chen and von Mikecz 2005, Li et al. 2003, Geiser et al. 2005). The presence of nanomaterials in both mitochondria and the nucleus opens the possibility for oxidative stress mediated indirect genotoxicity, and direct interaction of nanoparticles with DNA and histones. Besides oxidative stress, additional mechanisms of genotoxicity which may be specific for nanomaterials also need to be considered, such as possible mechanical interferences during cell division, and other sources of genotoxic effects (i.e. metal release by nanomaterials) (Gonzalez et al. 2008).

Several studies with nanoparticles have indicated that some nanoparticles may be genotoxic (reviewed by Gonzalez et al. 2008, Landsiedel et al. 2008). The most frequently used test was the Comet assay demonstrating the presence of DNA damage. For several nanomaterials a positive outcome on genotoxicity was observed including C60 fullerene, single walled carbon nanotubes (SWCNT), nanoparticles of cobalt chrome (CoCr) alloy, TiO2, nanosized metal oxide V2O3, Carbon Black (CB), and nanosized diesel exhaust particles.

The second most frequently used assay was the micronucleus assay in which the presence of micronuclei in dividing cells is indicative for chromosomal aberrations. In the micronucleus assay positive results were obtained for nanoformulations of TiO2, SiO2, CoCr, ZnO and multi-walled carbon nanotubes (MWCNT). For the gene mutation assays some studies showed a positive result for several nanomaterials including nano-FePt, SiO2, TiO2, MWCNT, and CB.

For all three assay systems used (Comet, micronucleus and gene mutation), negative results were obtained for TiO2, CB, SiO2, and single walled carbon nanotubes, while for some nanomaterials contrasting results were obtained (Landsiedel et al. 2008). The interpretation of the data presented in the reviewed papers was hampered by various limitations including the differences in the methodology used within one assay type, the use of non-standardised methods with different primary cells or cell lines, and by the sometimes minimal characterisation of the nanoparticles tested and the lack of information on possible contaminants.

For TiO2 and CB it was reported that the smaller (~20 nm) particles induced DNA damage while larger particles (~200 nm) did not (Gurr et al. 2005, Mroz et al. 2008, Rahman et al. 2002). Cobalt nanoparticles have been shown to induce more DNA damage than micron sized particles using human fibroblasts in tissue culture in the alkaline comet assay (Papageorgiou et al. 2007). In the micronucleus assay Co nanoparticles showed minor changes, whereas in the Comet assay for the same Co nanoparticles, clear statistically significant positive results were observed (Colognato et al. 2008).

Some studies showed that highly purified amorphous silica, with a low surface reactivity, was negative in the Comet assay (Barnes et al. 2008). This might suggest that nanoparticles with low surface reactivity are likely to be less genotoxic than others. In addition to the negative results of Barnes et al. (2008), very mild positive results (Yang et al. 2009) on DNA damage in Comet assay were reported after exposure of mouse embryo fibrolasts 3T3 to different concentrations of SiO2 nanoparticles (size 20 – 400 nm). In two types of genotoxicity assays i.e. the micronucleus assay and the gene mutation assay positive results were observed for silica nanoparticles (Landsiedel et al. 2008) while in the Comet assay weak positive results were obsreved (Yang et al. 2009).

A variety of genotoxicity (Ames test, clastogenicity in mammalian cells) and photo- genotoxicity (Photo-Ames test, photo-clastogenicity in mammalian cells) tests have been performed under GLP conditions on 14 different sunscreen-grade TiO2 (anatase and rutile; coated an uncoated; particle size range 11-60nm + one pigment grade – 200 000 nm). All results were negative. They were provided as an unpublished industry safety dossier but reviewed, summarised and published in the Opinion of the Scientific Committee on Cosmetic Products and Non-Food Products Intended for Consumers Concerning Titanium Dioxide (SCCNFP, 2000). Negative photo-clastogenic results were also found in chromosome aberration tests on Chinese hamster ovary cells with a variety of TiO2 particles (anatase, rutile; particles size: 14-60 nm) (Theogaraj et al. 2007).

However, others (Rahman et al. 2002, Wang et al. 2007) documented that ultrafine TiO2 particles increased the number of micronuclei in Syrian hamster embryo cells and a human B-cell lymphoblastoma (WIL2-NS) cells. In the latter model mutation frequency was increased in the HPRT test and DNA damage was indicated by the Comet assay. Positive results in micronucleus test and oxidative DNA damage were found recently in fish cell lines derived from rainbow trout and goldfish skin (Reeves et al 2008, Vevers and Jha 2008). It was suggested that several types of TiO2 (anatase; particle size of 255 – 420 nm) were not genotoxic but photo-genotoxic in mouse lymphoma and Chinese hamster lung cells (Nakagawa et al. 1997). This was further supported by the study of Dunford et al. (1997) which showed DNA oxidative damage in human fibroblasts (MRC-5) using the Comet assay. Similar to silica (SiO2) positive results were observed for TiO2 in all three types of genotoxicity assays (Landsiedel et al. 2008).

Inconsistent results were published for the genotoxicity of zinc oxide nanoparticles (Brayner 2008, Dufour et al. 2006, Nohynek et al. 2007, SCCNFP 2003, Yang et al. 2009). It was shown that silver nanoparticles (ca. 60 nm size) did not increase micronuclei formation in rat bone marrow in an in vivo 28-days oral exposure test in doses up to 1000 mg/kg (Kim et al. 2008).

The genotoxicity of C60 has been relatively well studied, but conclusions are conflicting. Dhawan et al. (2006) demonstrated a weak genotoxicity of colloidal C60 in Comet assay performed on human lymphocytes. Several types of carbon nanomaterials such as carbon nanoparticles that generate reactive oxygen species (ROS) were found to be genotoxic (Jacobsen et al. 2008). On the other hand, Zakharenko et al. (1997) reported no genotoxicity of C60 in concentrations as high as 450 μg/l in SOS Chromotest and slight genotoxicity (at 2.2 mg/l) when somatic mutation and recombination test in Drosophila melanogaster was used. Mori et al. (2006) obtained negative results for fullerenes in both a bacterial Ames test and mammalian cell chromosomal aberration tests.

In order to consider a specific engineered nanomaterial genotoxic, confirmation by two independent laboratories and in two test systems is necessary and minimal criteria should be met as proposed by Gonzalez et al. (2008). There are various inconsistencies in the results reported so far (reviewed by Gonzalez et al. 2008, Landsiedel et al. 2008). A major limitation in concluding whether a certain nanomaterial is genotoxic or not is the scarce description and minimal characterisation of the nanomaterial samples used in the various studies.

In conclusion, for some manufactured nanomaterials, in vitro genotoxic activity has been reported, but negative contradictory results were also obtained, and not all results could be confirmed by in vivo testing. One potential cause of inconsistencies is the difficulty in delivering the nanomaterials to the test systems appropriately. Most available in vitro/in vivo genotoxicity studies have been performed at high particle concentrations. In in vivo situations, this may be associated with marked inflammatory and proliferative responses, and hence may obscure and/or modify genotoxicity and even carcinogenicity readouts. In addition, various assays with different primary cells and cell lines were used which did not always show consistent results. Such inconsistencies may depend on physico- chemical characteristics of the test material investigated such as size, shape, aggregation/agglomeration state, surface properties, contaminants present and the cell type used.

5.5 Do nanoparticles have effects on the heart and blood vessels?

Air pollution is increasingly recognised as an important factor for cardiovascular disease in urban communities (reviewed by Mills et al. 2009). The occurring cardiovascular events including myocardial infarction and heart failure were attributed to the exposure to combustion derived nanoparticles that incorporate reactive organic and transition metal components. It was suggested that after inhalation the resulting respiratory inflammation induces systemic effects either directly by translocation from the lung or indirectly by yet unknown mediators (Mills et al. 2009). In view of these findings also for manufactured nanomaterials a risk for interaction with the cardiovascular system can be imagined.

Varous types of carbon nanomaterials were investigated and compared to a standard urban particulate matter by Radomski et al. (2005). Mixed carbon nanoparticles, single wall carbon nanotubes, multi wall carbon nanotubes, but not C60 fullerenes were found to stimulate platelet aggregation in vitro, and to accelerate vascular thrombosis in a ferric chloride model of thrombosis in a specific rat model. No information was presented on the characteristics of the carbon nanomaterials used, which limits the value of the observations. In a recent study, modified fullerenes (C60(OH)24) were found to facilitate adenosine diphosphate (ADP)-induced platelet aggregaton in vitro, whereas C60(OH)24 alone or carbon black did not (Niwa and Iwai 2007).

In contrast for several nanomaterials designed for drug delivery purposes, no or limited effects on platelet function in vitro was noted including alcohol/polysorbate nanoparticles (Koziara et al. 2005), gadolinium nanoparticles (Oyewumi et al. 2004), and nanostructured silica hydroxyethyl methacrylate biocomposites (Liu et al. 2008). Based on the observations some concern exists on the possible effect of manufactured nanoparticles on the cardiovascular system. However, so far this has not been clearly demonstrated to be the case for manufactured nanoparticles as well. Overall the information on the possible hazard of nanoparticles for cardiovascular effects is rather limited and needs expansion.