© 2003 Robert A. Freitas Jr. All Rights Reserved.

Robert A. Freitas Jr., Nanomedicine, Volume IIA: Biocompatibility, Landes Bioscience, Georgetown, TX, 2003

15.3.2.1 Pure Carbon Fullerenes and Nanotubes

Because the fullerenes are condensed ring aromatic compounds with extended pi systems, concerns about their possible carcinogenic properties have been raised from time to time [698, 917]. In regard to carbon nanotubes, inhalation toxicity has been the initial concern [669, 6060, 6061]. That’s because carbon nanotubes are rigid cylinders >1 nm wide and up to 100 microns long that crudely resemble the shape of asbestos fibers that have been linked to cancer. The dangers of asbestos first emerged in the early 1960s, when studies linked exposure to these silicate fibers with mesothelioma – a rare cancer of the lining of the chest or abdomen (pleura, pericardium, or peritoneum) that’s commonly fatal [669]. Asbestos fibers are small enough to be inhaled into the deep lung. Once embedded there, metals in the silicate fibers act as catalysts to create reactive oxygen compounds that go on to damage DNA and other vital cellular components. Asbestos expert Art Langer at the City University of New York’s Brooklyn College has worried that carbon nanotubes may “reproduce properties [in asbestos] that we consider to be biologically relevant.” Most notably, nanotubes are the right size to be inhaled. Their chemical stability means they are unlikely to be broken down very quickly by cells (hence could persist in the body), and their needlelike shape could damage tissue [669]. Morgan [6060] adds: “...the presence of long, extremely durable fibers in the lungs is worrisome. Nanotube ropes may indeed act like asbestos, and certainly if the nanotubes are wrapped up in something like PMMA they have a reasonably good chance of being damaging to the lung in moderate doses. But even if nanotubes are not damaging to the lung, we can’t presently exclude a transport mechanism to a more sensitive organ elsewhere in the body. It makes sense to start the study of fullerene toxicology through inhalation studies. Concurrently, to get a better understanding of the mechanism of injury, we could study things like fiber durability in the lung, and lung cell protein production and macrophage sensitivity.” In August 2002, Morgan [6060] announced that “we’ve recently come to an agreement with Dr. Joseph Brain, Harvard Department of Environmental Health, director of the Physiology program. He has agreed to study macrophage sensitivity and lung cell protein production in a mouse model, and will use nanotube material supplied by Dr. Edwards.”

Chunming Niu, a chemist with Hyperion Catalysis International (Cambridge MA) that produces 300 kg/day of multiwalled carbon nanotubes (MWNTs), admits that this “certainly is a concern. We treat our nanotubes as highly toxic material.” Rather than shipping nanotubes in powder form, Hyperion first incorporates the nanotubes into a plastic composite so that they cannot be inhaled [669]. As a precaution, many researchers who use carbon nanotubes in their work wear masks during procedures that could generate airborne plumes of the material [5227].

Brooke Mossman, a pathologist at the University of Vermont College of Medicine, is reported as arguing that it is asbestos’ ability to generate reactive oxygen compounds that makes it carcinogenic. Mossman says that the graphitic carbon structure of nanotubes is unlikely to react with cellular components to produce damaging byproducts: “We’ve worked with a lot of carbon-based fibers and powders and not seen any problems” [669]. In 2001, Huczko et al [2599] at the Medical University of Warsaw conducted preliminary experiments to explore whether carbon nanotubes act in lung tissue the way asbestos does. Four weeks after subjecting guinea pigs to soot that did or didn’t contain carbon nanotubes, pulmonary-function tests and inflammatory reactions (upon autopsy) were substantially identical between the groups. This led the researchers to conclude that “working with soot containing carbon nanotubes is unlikely to be associated with any health risk.” Huczko [5655] also provides evidence that fullerenes have minimal risk of allergic reaction or skin irritation. However, Huczko’s study [2599] evidently was performed without adequate controls and used techniques that have been abandoned by the EPA as not effectively evaluating the relevant criteria. Silvana Fiorito at the University of Montpellier in France found that 1-micron graphite particles stimulated rat cells to produce NO (nitric oxide, an indicator of immune response), but that neither carbon nanotubes nor fullerene cage molecules elicited NO production from these cells [5227]. Richard Smalley, the Rice University Nobelist and founder of Carbon Nanotechnologies Inc., was quoted in September 2002 as saying that an as-yet unreleased NASA study showed little cause for alarm over the biological safety of carbon nanotubes, though evidently one mouse tested died after receiving “vast amounts” of nanotubes in its lungs [5716].

K.D. Ausman [personal communication, November 2002], Executive Director of Rice University’s Center for Biological and Environmental Nanotechnology (CBEN), notes that in late 2002 the nanotube/asbestos comparison was still very much an open question, but that at least partial answers might be coming soon. Ausman notes that there appear to be two primary asbestos toxicity mechanisms.

The first toxicity mechanism involves the size and shape of the fibers, which prevents macrophages from engulfing the fibers and elicits a biochemical cascade that triggers the formation of scar tissue in the lungs. Given the current definitions of exposure limits (which include clearance rate in the denominator), no “safe” exposure limit can be set because these fibers are never cleared from the body. Says Ausman: “While nanotubes are much more rigid than asbestos fibers when normalized to aspect ratio, in practice they are not at all rigid individually because their aspect ratio is huge – witness the remarkable curvature present in typical buckypapers. However, the tubes are frequently present as bundles or, depending on the sample, as multiwalled tubes. In those cases, the rigidity may be sufficient to see similar problems as seen with asbestos. The jury is still out, although there is a paper currently in peer review that sheds some light on this question.”

The second toxicity mechanism of asbestos is due to very small amounts of bioavailable iron in the fibers which induces a type of cancer [5793, 5794]. The most carcinogenic forms of asbestos, crocidolite and amosite, contain up to 27% iron by weight as part of their crystal structure, and these minerals can acquire more iron after being inhaled, forming asbestos bodies [5794]. Ausman notes that while nanotubes themselves are unlikely to produce a similar effect, nanotubes are traditionally made from metal catalysts, in many cases iron. “In typical samples, the metal content is huge compared to that of asbestos,” he notes. “It is unknown how bioavailable it is, but again a paper that is in peer review should shed some light on this question.”

The results of two studies – the NASA study and the DuPont study – focused on single-walled carbon nanotubes (SWNTs) were announced at an American Chemical Society meeting [6212, 6213] in March 2003 as this book was going to press. According to one account [6212], the NASA team [6214] exposed groups of mice to one of four substances: (1) newly made SWNTs mixed with tiny grains of the metal catalyst used in making the nanotubes, (2) SWNTs treated to remove the metals, (3) carbon black, or (4) quartz nanoparticles having well-characterized toxicity. The mouse lungs were instilled with a solution containing either 0.1 or 0.5 micrograms of material suspended in inactivated mouse serum. After 90 days, standard histological tests showed that all the particles made their way into the alveoli and most remained there intact. Even at the lower dose, the nanotubes (with or without metal particles) triggered granuloma formation surrounding the material, “a significant sign of toxicity.” By comparison, the carbon black particles triggered little inflammation. In the second study at DuPont [6215], SWNT-induced granuloma formation was similarly observed but the inflammation appeared to tail off after 90 days, and the group concluded that nanotubes were less toxic than the quartz dust. Both groups cautioned that conclusions about nanotube toxicity must wait until researchers learn more about how the animals’ lung tissue reacts to airborne SWNT particles [6248].

Almost all of the experimental studies to date have focused on the biocompatibility of C₆₀ and related spheroidal fullerenes, rather than nanotubes. C₆₀ (though easily destroyed by O₃ in the air even when shielded from UV and visible light [5508]) is present naturally in the environment [5508-5512, 6055-6058], being generated in trace amounts in virtually any sooty smoky flame, such as in forest fires, campfires and candle flames, and C₆₀ has been discovered in meteorites [5509], in space [6056], and in carbon-rich shungite (a metamorphic Precambrian rock) [5510, 6058], in more ancient formations dated back 1.85 billion years [5511], and in rocks at the Cretaceous-Tertiary boundary [5512, 6057]. (Carbon nanotubes might yet be found in nature [6059].) Natural fullerene “concentrations have been exceedingly low and dose to humans, if any, have probably been trivial” [6060]. Industrial toxicology reports show that pure C₆₀ has virtually no inflammatory effect in mice and rats and does not elicit an immune response [698]. However, fullerenes readily adsorb molecular oxygen from air [1604]. Soon after bulk quantities of fullerenes became available for laboratory experimentation in the early 1990s, it was found that in the presence of light and oxygen, the C₆₀ molecule could pass its superfluous excitation energy onto nearby oxygen molecules, creating a long-lived but very reactive form of oxygen called singlet oxygen [680, 681, 918, 919]. It was quickly suggested that this could present potential health risks [1099]. Pure C₆₀ is a singlet oxygen-generating agent. It yields singlet oxygen in very high amounts and is completely inert to photo-oxidative destruction [682]. One early experiment with macrophages observed little influence on the formation of reactive oxygen species by C₆₀ but found that raw soot from fullerene production was oxidatively active with cells under the influence of light, though not cytotoxic [683]. (It also appears that C₆₀ traps the chemical contaminant naphthalene much differently than soil [6255], and Tomson et al [6256] are studying the impact and transport of carbon nanomaterials in the environment.)

Other experiments have since shown that C₆₀ efficiently generates singlet oxygen when irradiated with light [684]. For example, when C₆₀ was incorporated into rat liver microsomes in the form of its cyclodextrin complex and exposed to UV or visible light, it induced significant oxidative damage to lipids (e.g., lipid peroxidation as assayed by thiobarbituric acid reactive substances, lipid hydroperoxides, and conjugated dienes) and to proteins (e.g., assayed by protein carbonyls and loss of membrane-bound enzymes), predominantly due to the production of singlet oxygen [685]. Exposing C₆₀ to laser pulses at 355 nm or 532 nm in the presence of O₂ produces large quantities of singlet oxygen. This occurs not by chemical reaction but by an energy transfer from the highly populated C₆₀ triplet state to molecular oxygen [919]. K.D. Ausman notes that as the degree of derivitization increases for functionalized C₆₀, the efficiency of singlet oxygen sensitization goes down because the 1O2 sensitization goes through the C₆₀ triplet state, and both the absorption cross-section at relevant wavelengths and the quantum yield of conversion to T1 drops. Light emission from carbon nanotubes was claimed to be observed in aromatic amine solvents [4635], and Weisman and Smalley et al [5795] have found near-infrared emission from SWNTs. Photoinduced biochemical activity* has also been reported in fullerene carboxylic acid [922]; a “teflon ponytail” fullerene is an efficient sensitizer for (increasing) singlet oxygen formation in fluorous solvents [4633]; and porphyrin-fullerene hybrids have been synthesized [4634].

* Some specific samples of single-walled carbon nanotubes (SWNTs) in air have emitted a loud pop and suddenly burst into flames when exposed to the light from an ordinary camera flash [4697]. According to Pulickel Ajayan of Rensselaer Polytechnic Institute, the initial popping noise is generated by the heating of the oxygen inside and between the nanotubes, which produces a shock wave that causes the carbon to oxidize, sparking combustion, when the temperature reaches 600-700 °C. The explosion occurs because the black carbon nanotubes absorb light so efficiently that, when it is converted into heat, the heat cannot dissipate quickly enough across bunched-up tubes. MWNTs, unlike SWNTs, do not explode, and other nanotube researchers have taken flash photographs of SWNTs without triggering combustion.

On the other hand, when not in an excited state, C₆₀ does not react with singlet molecular oxygen but quenches it slowly with a rate constant of k_q ~ 5 x 10⁵ / M-sec [919]. It is claimed that a single C₆₀ molecule can absorb dozens of these reactive chemical species [680]. Water-soluble peptide (DL-alanine) and dipeptide (DL-alanyl-DL-alanine) derivatives of C₆₀ are also known to quench pyrene fluorescence and erythrosine phosphorescence, both in water solution and in liposomes [686]. Charge transfer across C₆₀-doped bilayer lipid membranes has also been investigated by cyclic voltammetry to evaluate membrane suitability in practical biosensors [687], and C₇₀ can act both as a photosensitizer for electron transfer from a donor molecule and a mediator from electron transport across a lipid bilayer membrane [688]. Indeed, it was found that the steady-state photocurrent density obtained from the C₇₀-bilayer system was about 40 times higher, at comparable light intensities, than that of the carotene-porphyrinquinone system, previously the most efficient artificial system known in the early 1990s. The C₇₀-bilayer system has a quantum yield of about 0.04, while the stability (tens of minutes) and turnover number (10³ electrons transported per C₇₀ before decay) are 1-3 orders of magnitude greater than in other systems [688].

Can fullerenes spontaneously traverse lipid bilayers? A simple C₆₀ cage easily accepts electrons, acquiring a negative charge [680], and nanotubes readily pick up negative charges in aqueous suspension [689, 690]. The large electron affinity of fullerenes like C₆₀ or C₇₀ is well known [691]. Fullerenes exhibit enhanced electron-withdrawing character upon increase of their molecular size [691]. The negative charge of lipid bilayers (and most in vivo biological surfaces) might argue for fullerene objects to be slightly repelled from cell membranes. However, E. Pinkhassik notes that the negative charge of fullerenes is delocalized over a very wide surface and therefore should not be the decisive factor.

In general, hydrophobic molecules readily insert into the interior of the bilayer membrane, so we should expect fullerenes to insert into bilayers as well. Indeed, organic nanotubes with hydrophobic groups on their exterior surface are observed to spontaneously insert into lipid bilayers [692], and C₆₀ is highly hydrophobic [693, 725]. Carboxylic acid C₆₀ derivatives having polar character can readily enter lipid membranes [745], and water-solubilized peptide (DL-alanine) and dipeptide (DL-alanyl-DL-alanine) derivatives of C₆₀ can localize inside an artificial membrane, penetrate through the lipid bilayer of phosphatidylcholine liposomes, and perform activated transmembrane transport of bivalent metal ions [686]. The rapid uptake of radiolabeled C₆₀ into human cells (~50% of C₆₀ present in serum, within 6 hours [695]) does not result in acute toxicity and does not affect the proliferation rate of human keratinocytes or fibroblasts [694, 695]. Open-ended carbon nanotubes that spontaneously inserted into cell membranes could promote cell lysis much like porins (Section 10.4.1.4), transmembrane siphons (Section 10.4.2.1), or the membrane attack complex (Section 15.2.3.2). Additional research will be required to identify all the parameters which may govern the spontaneous insertion of fullerenes into cell membranes.

E. Pinkhassik notes that the insertion of fullerenes into membranes can be directly relevant to nanorobot construction if one considers the danger of fullerene-based nanorobot appendages poking into cells. Accidental whole-nanorobot diffusion through bilayers is unlikely due to the large size of such a device, but an individual nanorobotic arm or its protrusions might potentially spontaneously enter the membrane of some cells. This issue could arise for graphene-based appendages or for any other hydrophobic material used in the construction of a small-diameter nanorobotic arm whose feature lengths exceed a half-membrane thickness, or about 3-5 nm (Section 8.5.3.2).

Another potential nanomedical concern is carcinogenicity [698, 917]. Many organic substances that have aromatic ring systems, such as benzene, are carcinogens because a conjugated carbon ring has the appropriate size and shape to be intercalated into DNA, thus promoting cancer. But buckyballs appear to be too big and round to be incorporated into DNA [680], as are buckytubes (essentially a curved array of such rings). So these should not present a problem as long as they remain intact. Preliminary experiments with mouse skin exposed to pure C₆₀ and C₇₀ confirm this expectation [698], though more data is needed to increase confidence in the lack of carcinogenicity. The possible carcinogenic risks of nanotubes was discussed above.

Genotoxicity is defined as a serious injury to the chromosomes of the cell, such that when the cell divides, fragments of chromosomes and micronuclei remain in the cytoplasm. Experiments by Zakharenko and colleagues in 1994 [696] and in 1997 [697] examined the genotoxicity of C₆₀ in prokaryotic cells (E. coli) and in eukaryotic cells (Drosophila somatic wing cells). No genotoxicity was observed at a C₆₀ concentration of 0.45 µg/cm³ in any of the cells, but at the highest fullerene concentration of 2.24 µg/cm³ a slight genotoxic effect was observed in the eukaryotic cells.

A related concern is mutagenicity – the production of coding flaws in chromosomes that otherwise may retain much genetic functionality. Miyata et al [681, 1092] found that C₆₀ dissolved in polyvinylpyrrolidone was mutagenic for several Salmonella strains in the presence of rat liver microsomes when irradiated by visible light. Their results suggested that singlet oxygen was generated and that the mutagenicity was caused by the indirect action of singlet oxygen producing phospholipid peroxidation (principally of the linoleate fraction) in rat liver microsomes, leading to oxidative DNA damage (probably with the generation of radicals at the guanine bases only). However, a confusing factor in this study was their use of polyvinylpyrrolidone, a solvent known to cause liver cancer. Indeed, many studies of the biocompatibility of pure fullerenes have had to employ biologically harmful solvents, since naked fullerenes are not soluble in physiological saline. Teratogenicity (e.g., fetotoxicity; Section 15.2.8) of pure fullerenes has yet to be seriously investigated.

Cell and tissue biocompatibility experiments on pure fullerenes – principally C₆₀ – have begun. In many situations, pure fullerenes are almost completely bioinactive [5230]. For example, the dermal toxicity of pure C₆₀ was studied by applying a solution of C₆₀ in benzene to mouse skin epidermis. A 200 mg topical dose produced no acute toxic effect on either DNA synthesis or ornithine decarboxylase activity over a 72-hour time course after treatment [698]. Repeated epidermal administration of fullerenes for up to 24 weeks (after initiation with a polycyclic aromatic hydrocarbon or PAH) resulted in neither benign nor malignant tumor formation. Promotion with a phorbol ester used as a positive control resulted in the formation of benign skin tumors [698]. In another study of the pharmacological effects of fullerenes on various tissues, pure C₆₀ was applied to guinea pig trachea, right atria, ileum, and stomach (fundus) tissues, and to rat vas deferens and uterus [699]. A 4 µM (~3 µg/cm³) dose had no direct effect in any tissue. A short-term repeated application of a 30 mg/kg dose of C₆₀ for 4 weeks significantly reduced the potencies of acetylcholine in guinea pig ileum and its longitudinal muscle. C₆₀ was found to have no direct or antagonistic properties toward drug receptors, though sub-chronic exposure decreased responsiveness. No effect on bacterial growth rates was found in 22 microbial strains exposed to C₆₀ at doses of 43.2 µg/cm³, and there was no cytotoxicity to human macrophage, leukocyte, or monocyte [2383]. (Therapeutic doses in rodent models are typically measured in µg.)

Besides the aforementioned work of Fiorito [5227], by early 2002 the only direct study on carbon nanotube cytocompatibility was by Mattson et al [4820] who grew embryonic rat-brain neurons on multiwalled carbon nanotubes. They reported that on unmodified nanotubes, neurons extend only one or two neurites, which exhibit very few branches. In contrast, neurons grown on nanotubes coated with 4-hydroxynonenal (a bioactive molecule) elaborate multiple neurites which exhibit extensive branching [4820]. This result was said to “establish the feasibility of using nanotubes as substrates for nerve cell growth and as probes of neuronal function at the nanometer scale.”

Other experiments have found some bioactivity [5233], though usually only in functionalized fullerenes (Section 15.3.2.2). For instance, C₆₀ solubilized with polyvinylpyrrolidone (PVP) in water was applied to the rat limb bud cell differentiation system and very strongly promoted cell differentiation (up to a 3.2-fold increase) [700]. PVP alone inhibited the cell differentiation in proportion to its concentration, suggesting that a specific promoting action on chondrogenesis may exist for C₆₀. In a test of the immune reactions of macrophages, raw soot from fullerene production and purified C₆₀ were incubated with alveolar macrophages and macrophage-like cells. The effects of this treatment were compared to DQ12 quartz which is known to damage BAM and HL60M macrophages. Neither soot nor C₆₀ were cytotoxic compared to quartz, but C₆₀ did induce some chemotactic activity, although less than the soot or the quartz [683]. Nobuhisa Iwata et al [701] investigated the effects of C₆₀ on the activities of glutathione S-transferase (GST), glutathione peroxidase (GSH-Px), and glutathione reductase (GR) enzymes in rodent and human liver. C₆₀ inhibited GST activity toward trans-4-phenyl-3-buten-2-one in rat liver and toward ethacrynic acid in mouse liver, while activity toward other substrates was not affected. In human liver, C₆₀ again inhibited GST activity toward ethacrynic acid and moderately inhibited GSH-Px and GR activities as well. Lin and Wu [702] conducted a study of platelet activation, using amine-terminated silane coupling agents to graft C₆₀ molecules onto a polyurethane (PU) surface pretreated with oxygen plasma activation. Electron spectroscopy for chemical analysis (ESCA) analysis showed that the C₆₀ molecules spontaneously attached via nucleophilic additions to the fullerene double bonds which fuse two six-membered rings. In vitro platelet adhesion assay subsequently demonstrated that the C₆₀-graft-PU activated more platelets than the nontreated PU control, though the researchers [702] admitted this might be due to “the few residual amine functional groups which are left over after the C₆₀ grafting reaction.”

Pure fullerenes are fairly chemically inert. They are stable substances in air or in solution and can be purified by sublimation without decomposition [678]. Unmodified fullerenes are virtually insoluble in water [2566], suggesting a low reactivity with biological tissue [680]. They are only slightly soluble in ethanol but are much more soluble in aromatic solvents and in carbon disulfide [695, 910, 920, 921], with highest solubility in 1,2-dichlorobenzene [2552, 2566]. Mustard-yellow aqueous suspensions of C₆₀ have been prepared that do not settle out upon standing for more than 3 months. SEM examination of these suspensions shows the suspended particles are spherical clumps of C₆₀ between 250-350 nm in diameter. There is some evidence for suspended-particle oxidation, to C₆₀O, after prolonged storage in air [695]. Pure C₆₀ and C₇₀ have also been solubilized by encapsulation inside hollow aggregates of block copolymers [2547].

Intact pure fullerene surfaces are unlikely to be attacked by chemicals naturally present in the human body, although the possibility of a graphene surface being chemically attacked by short-lived reactive intermediates that form during metabolic processes cannot be entirely ruled out. There is just one unconfirmed (but somewhat dubious) report [2384] of the metabolization of C₆₀ by a microbe – it is claimed that several species of fungi can grow in fullerenes as their sole source of carbon. Fullerenes can have numerous chemical groups attached to their surfaces via processes known variously as solubilization, conjugation, complexing, derivatization, or functionalization, typically using intermediary reactive chemical species not normally found inside the human body. Examples of chemical groups that have been added to C₆₀ include hydroxyls (making water-soluble fullerols or fullerenols) [703-705], carboxylic acid [922], proline [706], polyamines [707], amine-terminated silane [702], aldehydes [708], pyrrolidines [709, 710] and poly(vinylpyrrolidone) [921], polyethylene glycol [684], cyclodextrin [909], cyclopropane [711], lipid micelles and vesicles [910], methanofullerenes [711] and lipidized methanofullerenes [712], and even proteins [725, 911] making water-soluble “fullereyl protein” [724, 911]. The C₆₀ cage can reversibly accept up to 6 electrons under suitable conditions [1100]. Carbon nanotubes have been derivatized with, among other chemical groups, benzyne [713, 714], thionychloride and octadecylamine [715]. The variety of chemical modifications of fullerenes may warrant the study of interactions between the fullerene surface and the reactive species found in the human body.

With proper chemical treatment, C₆₀ can have a stable orifice of fixed diameter opened up in its side, only allowing atoms smaller than a certain size to enter [716, 717]. The open ends of ruptured carbon nanotubes, cut fullerene pipes, or ruptured fullerene surfaces would most likely be terminated with carboxylic acid groups if the cutting occurred in an acidic environment [718]. In non-acidic environments, hydroxyls, amines (basic), hydrocarbons (hydrophobic), or other terminating moieties could be present instead, producing alternative chemical reactivities near the break site.

C₆₀ remains structurally intact (~99%) when exposed to a neutron flux of up to 1.5 x 10¹⁶ neutrons/m²-sec [1089].

Last updated on 30 April 2004