Differences and similarities between carbon nanotubes and asbestos fibers during mesothelial carcinogenesis: Shedding light on fiber entry mechanism

Abstract

The emergence of nanotechnology represents an important milestone, as it opens the way to a broad spectrum of applications for nanomaterials in the fields of engineering, industry and medicine. One example of nanomaterials that have the potential for widespread use is carbon nanotubes, which have a tubular structure made of graphene sheets. However, there have been concerns that they may pose a potential health risk due to their similarities to asbestos, namely their high biopersistence and needle‐like structure. We recently found that despite these similarities, carbon nanotubes and asbestos differ in certain aspects, such as their mechanism of entry into mesothelial cells. In the study, we showed that non‐functionalized, multi‐walled carbon nanotubes enter mesothelial cells by directly piercing through the cell membrane in a diameter‐ and rigidity‐dependent manner, whereas asbestos mainly enters these cells through the process of endocytosis, which is independent of fiber diameter. In this review, we discuss the key differences, as well as similarities, between asbestos fibers and carbon nanotubes. We also summarize previous reports regarding the mechanism of carbon nanotube entry into non‐phagocytic cells. As the entry of fibers into mesothelial cells is a crucial step in mesothelial carcinogenesis, we believe that a comprehensive study on the differences by which carbon nanotubes and asbestos fibers enter into non‐phagocytic cells will provide important clues for the safer manufacture of carbon nanotubes through strict regulation on fiber characteristics, such as diameter, surface properties, length and rigidity. (Cancer Sci, doi: 10.1111/j.1349‐7006.2012.02326.x, 2012)

Nanotechnology is an emerging technology with the potential to make our daily lives better, as long as we know how it works.1 Asbestos, a natural fibrous silicate mineral, was once hailed as a miraculous stone and used throughout the world during the 20th century; however, it is now referred to as a time bomb due to its carcinogenicity through inhalation.2 The mechanism of asbestos‐induced carcinogenesis is yet to be fully unraveled, but it is currently known that asbestos fibers' physical characteristics, especially high biopersistence and needle‐like structure, are among the crucial factors responsible for causing cancer.3, 4 Carbon nanotubes are a nanomaterial that was recently discovered, manufactured, and commercially distributed. These tube‐shaped, carbon nanomaterials have raised social concerns due to their strong resemblance to asbestos fibers, especially in regard to the two features mentioned above.

The toxicological similarity of carbon nanotubes and asbestos fibers was reported by Poland et al.5 They reported that long fibers, whether they are carbon nanotubes or asbestos fibers, evade complete phagocytosis by macrophages and cause inflammation. This failure in phagocytosis, termed frustrated phagocytosis, has gained attention as a potential mechanism to explain the inflammogenicity and carcinogenicity of fibers.6, 7, 8

However, despite their similarities in certain aspects, carbon nanotubes and asbestos fibers vary in many physicochemical features, including constituent element, weight, and surface property. Therefore, we believe that these two distinct types of fibers do not necessarily exert toxicological effects through the same mechanism. One key difference that distinguishes these two fibers might be the mechanism of fiber entry into mesothelial cells. Considering that fiber entry into mesothelial cells is a key step in carcinogenesis, the difference between nanotubes and asbestos may provide a clue to aid in the development of safer carbon nanotubes. Moreover, the elucidation of the mechanism by which nanotubes enter into cells is of great use in both toxicology and medical applications.9

In this review, to establish a difference between the toxicological paradigms of carbon nanotubes and asbestos fibers, we summarize the current understanding of the differences and similarities between these two fibrous materials, with an emphasis on the entry mechanism of carbon nanotubes and asbestos fibers into non‐phagocytic cells.

Overview of Asbestos‐Induced Carcinogenesis

Before describing the similarities and differences between carbon nanotubes and asbestos fibers, we briefly outline the mechanism of asbestos‐induced carcinogenesis. Asbestos fibers are inhaled and eventually reach the pulmonary alveoli. During this process, asbestos fibers come into contact with epithelial cells in the trachea, bronchus and alveoli, as well as alveolar macrophages, leading to epithelial cell injury and macrophage activation. Small particles are well phagocytosed by macrophages and are excreted from the respiratory system via the lymphatic system. However, other long or large fibers that evade phagocytosis are not cleared and stay inside the lung for a long period of time, leading to chronic inflammation. Fibers that remain in the lung can physically penetrate through alveolar epithelial cells and visceral mesothelial cells and reach the parietal mesothelial cells due to negative pressure in the pleural cavity. Donaldson et al. hypothesized that long fibers (length > ~15 μm) tend to be trapped around the parietal stomata and the entryways to lymphatic vessels, as they cannot flow into the small openings like short fibers. Furthermore, they argued that mesothelial cell injury and macrophage activation occur at these sites of fiber accumulation, which then lead to the formation of mesothelioma.3

To date, we have four main criteria to assess toxicological and carcinogenic features of fibers: (i) translocation of fibers, such as from trachea to alveoli and from alveoli to parietal pleura; (ii) biopersistence, which determines the longevity of fibers inside organisms; (iii) mesothelial or epithelial cell injury caused by fibers; and (iv) activation of macrophages that phagocytosed the fibers. We will briefly review the translocation and biopersistence of fibers. Thereafter, we will provide an in‐depth discussion of the differences between carbon nanotubes and asbestos fibers in the mechanism of mesothelial or epithelial cell injury, with an emphasis on the cellular entry mechanism of these fibers. We will also compare the effects of carbon nanotubes and asbestos fibers on macrophage activation.

Translocation and Biopersistence of Fibers

Based on clinical observations, malignant mesothelioma develops most frequently at the parietal pleura, which unlike the visceral pleura, is not directly adjacent to the lung parenchyma. Appearance at the parietal pleura is reflected in the clinical stage. Such findings suggest that asbestos fibers somehow migrate through the lung parenchyma and reach the pleural cavity, where they come into contact with the parietal pleura. The exact mechanism by which fibers finally reach the parietal pleural has yet to be confirmed; however, Miserocchi et al.10 suggested that fibers in alveolar space travel physically through the paracellular pathway to reach the interstitial space between the lung parenchyma and visceral pleura. Then, the fibers are able to translocate to almost all organs via fluid flow, such as lymph and blood. Finally, some fibers get stuck in the parietal pleura, especially around the openings of the lymphatic stomata. An alternate hypothesis is that fibers directly translocate through the lung parenchyma and visceral pleura to reach the parietal mesothelium.11 Although the high physical resistance of the visceral pleura is difficult for any fiber to migrate through, it can still take place during chronic inflammation. During inflammation, the interstitial space exerts higher pressure, making it is easier for fibers to pass through the visceral pleura.10 Regardless of whether the fibers travel through the vascular system or directly translocate, they will be cleared by the lymphatic stomata when they reach the pleural cavity, and this pathway may partially explain why longer fibers (>8 μm), which are more likely to be trapped around the stomata compared to shorter fibers, are more carcinogenic in vivo.3, 12

The translocation of inhaled carbon nanotubes to subpleural tissue has been reported.13 The nanotubes used were suspended in phosphate‐buffered saline containing pluronic F‐68. As will be discussed later, the surface characteristics and size of nanotubes greatly affect their behavior. For example, carbon nanotubes that were covalently bonded with amino groups were easily distributed to each organ and ultimately excreted in the urine.14 This type of modification is called functionalization and is often performed to relieve the strong hydrophobicity of carbon nanotubes. The type of functional groups present on the surface of carbon nanotubes controls their hydrophilic/hydrophobic properties, which play important roles in the translocation, distribution and excretion of carbon nanotubes inside organisms. While this variable property of carbon nanotubes and its influence on cells is of interest, the biological behavior and effects of functionalized nanotubes should be carefully differentiated from those of pristine, or non‐functionalized nanotubes, which are the ones that are industrially produced, commercially distributed and of toxicological concerns. Despite the fact that functionalized nanotubes show little toxicity, it is currently unrealistic to functionalize high amounts of pristine nanotubes. Therefore, it is essential to carefully consider how pristine nanotubes would behave inside organisms for the toxicological assessment of carbon nanotubes.

Biopersistence is defined as the durability of materials inside organisms and is known to affect fiber translocation. Highly biopersistent fibers contribute to sustained inflammation and cell injury followed by the development of cancer.15 Clearance of fibers is affected by the dimension of the fiber, as long fibers that fail to be phagocytosed by macrophages are not carried and excreted.3 Furthermore, as mentioned above, long fibers are likely to be trapped at lymphatic stomata, leading to a failure of fiber clearance.12 In addition to the fiber dimension, the chemical composition of asbestos and carbon nanotubes renders them chemically stable, which enhances biopersistence because biodegradable fibers would be diminished following macrophage digestion. Mutlu et al.16 showed that carbon nanotubes that were well dispersed in pluronic solution were gradually cleared from the mouse lung. However, the translocation and biopersistence of “pristine” carbon nanotubes that are not coated by modifying materials should be investigated to gain further understanding of the toxicological property of carbon nanotubes.

Mesothelial and Epithelial Cell Injury Induced by Fibrous Materials

The carcinogenic feature of asbestos fibers and carbon nanotubes has gained a great deal of attention from both scientists and the public. Fibrous materials have the potential to transform normal cells, constituting a well‐organized tissue, into cancer cells by causing chromosomal aberrations and/or gene mutations that lead to invasion and destruction of the surrounding tissue. The target cells in fibrous material‐induced carcinogenesis are mesothelial cells or epithelial cells, but not macrophages. Therefore, the mechanism by which the DNA of mesothelial/epithelial cells is damaged has been of great interest. Currently, there are four key phenomena by which fibrous materials are postulated to cause DNA damage: (i) free radical generation on the surface; (ii) physical interaction between DNA and fibrous materials; (iii) molecule adsorption on the material surface; and (iv) chronic inflammation, as previously described.4 All of these mechanisms might take place simultaneously during fibrous material‐induced carcinogenesis, but the relative contribution of each may differ between carbon nanotubes and asbestos fibers. For example, carbon nanotubes piercing through the cellular and nuclear membrane might be more likely to directly associate with chromosomes than endocytosed asbestos fibers because penetrating carbon nanotubes are not covered by a membrane structure that would separate materials from the nuclear components (Fig. 1), as discussed previously.17

Figure 1.

Distinct uptake mechanism of asbestos fibers and carbon nanotubes by mesothelial cells. Chrysotile asbestos fibers (red arrows) are taken up by a mesothelial cell. Surrounding membranous structure around asbestos (blue arrowheads) indicates that the asbestos is endocytosed. On the other hand, a carbon nanotube (red arrows; diameter = ~50 nm) directly pierces the cellular and nuclear membrane. The human peritoneal mesothelial cell line was established as previously described.17 The cells were fixed after a 3 h‐incubation with indicated fibers. To increase the resolution, ultrathin sections were prepared at 80 nm thickness for asbestos. Considering the fact that carbon nanotubes are too rigid for cutting with diamond‐knife, the sections were prepared at 500 nm thickness for nanotube.

Non‐phagocytic cells, such as epithelial and mesothelial cells, form a continuous cell layer that acts as a barrier to cover and protect the underlying structure. Following exposure to fibrous materials, some of the cells undergo apoptosis or programmed necrosis, while others are able to evade such cell death mechanisms.18, 19, 20 The remaining live cells that hold these fibers inside are the most likely to become cancer cells after fibrous material‐induced DNA damage. As a consequence of DNA damage in mesothelial cells, homozygous deletion of the Cdkn2a/2b tumor suppressor genes is frequently observed, not only in asbestos‐induced mesothelioma21, 22 but also in carbon nanotube‐induced mesothelioma.17 We previously showed that mesothelioma induced by iron saccharate also displayed homozygous deletion of Cdkn2a/2b,23 suggesting that the loss of these genes might be a major event in mesothelial carcinogenesis.

Factors Involved in Carbon Nanotube and Asbestos Fiber Uptake by Non‐Phagocytic Cells

Although the exact mechanism of DNA damage is yet to be elucidated, uptake of fibrous material by non‐phagocytic cells appears an essential step in carcinogenesis. Furthermore, we believe that the key factor needed to differentiate carbon nanotube toxicology from that of asbestos fiber is the mechanism by which carbon nanotubes and asbestos fibers enter non‐phagocytic cells. As we have recently reported, carbon nanotubes directly pierce through the cell membrane to enter mesothelial cells in a diameter‐ and rigidity‐dependent manner, whereas asbestos fibers are actively endocytosed by these cells regardless of their diameter.17 This study was the first direct evidence describing the toxicological difference between carbon nanotubes and asbestos fibers regarding the mechanism of the entry of these materials into mesothelial cells. In this review, we summarize the previous literature describing how non‐phagocytic cells take up carbon nanotubes and asbestos fibers to refine and expand our understanding.

Many factors that are involved in the entry of fibrous materials into non‐phagocytic cells can be roughly divided into two categories: material characteristics and cell condition. Fibrous materials vary largely from each other in terms of their length, diameter, constituents, surface area, volume and weight. In addition to such physical variations, carbon nanotubes often contain contaminant metals because they are chemically synthesized in the presence of metal catalysts, while asbestos fibers are mined from mountains. It is also of note that the methods used to prepare suspensions of fibrous materials and the assays used for their characterization differ from lab to lab, which could account for differing results. For in vitro studies, fibrous materials are often suspended in solutions that may contain salt, proteins, DNA, polymers or alcohol, which could influence the results. Although there is a vehicle control group in cell‐based experiments, the possibility that these solvents might affect the interaction of cells with carbon nanotubes or asbestos fibers should be taken into account. Information regarding proteins or other molecules that are adsorbed onto fiber surfaces is described elsewhere and is known to affect fibrous material behavior and its effects on cells.24, 25

Cell type should also be carefully discussed in toxicology of asbestos fibers and carbon nanotubes. There are two primary cell types involved in this toxicology: non‐phagocytic cells and phagocytic cells. The interaction of each cell type with fibrous materials is important in this toxicology, but the cellular behavior and its role in pathophysiological conditions are very different.

Asbestos Fibers are Endocytosed by Mesothelial and Epithelial Cells

Because mesothelial and epithelial cells are not professional phagocytes, we should first address the question of whether asbestos fibers and carbon nanotubes enter into these cells when assessing the toxicity of these materials. The internalization of asbestos fibers into various non‐phagocytic cells has been reported by many researchers (Fig. 1). Suzuki and Jaurand et al. beautifully demonstrated that alveolar epithelial cells and pleural mesothelial cells internalized chrysotile asbestos into the phagolysosome.26, 27 Regarding the mechanism of the internalization, an interaction between the membrane proteins of the cells and proteins adsorbed onto the surface of asbestos fiber was investigated. Adsorption of vitronectin, a serum protein, onto the asbestos surface (chrysotile and crocidolite) was reported to enhance asbestos internalization by mesothelial and epithelial cells via the integrin alphaVbeta5 signaling pathways.28, 29, 30 Following the internalization of asbestos fibers into mesothelial cells, we have observed in the previous study that the asbestos fibers were surrounded by the small GTPase, Rab5a, which is a marker of the early endosome and phagosome.17, 31 Considering that the size of an endosome can be up to 1 μm,32, 33 asbestos fibers that are larger than a few microns could be internalized by a phagocytosis‐like mechanism rather than that of the endosome.

Cellular Uptake of Carbon Nanotubes Varies

Carbon nanotubes are divided into two groups depending on their structure: single‐walled carbon nanotubes (SWCNT) with a diameter of 1–3 nm and a length of 5–30 nm34 and multi‐walled carbon nanotubes (MWCNT) with a diameter of 3–200 nm and a length of tens of nanometers to microns.35, 36 As discussed previously, the size of carbon nanotubes is an important determinant of how the carbon nanotubes enter non‐phagocytic cells. Several reports have suggested that carbon nanotubes are able to enter non‐phagocytic cells via endocytosis or membrane piercing; however, a reasonable explanation on the difference has yet to be established. To identify common features of carbon nanotubes that determine the mechanism of cell entry, we have summarized the current literature describing carbon nanotube entry into non‐phagocytic cells, with a focus on length, diameter and surface functionalization of carbon nanotubes, cell type, entry mechanism and intracellular localization (Table 1).

Table 1.

Analysis of cellular entry of carbon nanotubes in non‐phagocytic cells in the literature

Nanotube	Length	Diameter	Surface	Cell type	Toxicity	Mechanism	Intracellular location	References
Endocytosis
SWCNT	~100–1000 nm	1–5 nm	Covalently functionalized (oxidized) + streptavidin	HL60 and Jurkat	Little toxicity	Endocytosis	Endocytic vesicles	51
SWCNT	Tens to hundreds of nm	~1.5 nm	Covalently functionalized (oxidized) + streptavidin, albumin, protein A or cytochrome c	HeLa, NIH3T3, HL60 and Jurkat	Little toxicity	Endocytosis	Endocytic vesicles	50
SWCNT	Short (145 nm) and long (1250 nm)	0.7–1.3 nm in ref.	Non‐covalent binding with Pluronic F127	HeLa	ND	Endocytosis	Perinuclear localization	63
MWCNT	7000–10 000 nm	80–130 nm	Non‐functionalized and suspended in PBS containing 0.1% gelatin	MESO‐1, BEAS‐2B and IMR‐32	Toxicity dependent on cell type	Endocytosis	Perinuclear localization without nuclear import	44
SWCNT	ND	1 nm in ref.	Non‐covalent binding with DNA	NIH3T3	Little toxicity	Endocytosis and exocytosis	Endocytic vesicles including lysosomes	64
SWCNT	~150 nm	~1.2 nm	Non‐covalent binding with DNA or phospholipid‐conjugated folic acid	HeLa	Selective toxicity by coupling near infrared light and interaction of folic acid‐nanotube and folate receptor‐expressing cancer cells	Folate receptor‐mediated endocytosis	ND	40
SWCNT	100–200 nm	1–3 nm	Covalently functionalized (oxidized) + chitosan, Alexa Fluor 488 and folic acid	HepG2	Little toxic up to 0.05 mg/mL	Folate receptor‐mediated endocytosis	Cytoplasm	42
SWCNT	100–300 nm	1–5 nm	Covalently functionalized (amino group, cisplatin and folic acid)	KB, JAR and Ntera‐2	Selective toxicity by coupling cisplatin and interaction of folic acid‐nanotube and folate receptor‐expressing cancer cells	Folate receptor‐mediated endocytosis	Endosomes	39
SWCNT	50–100 nm	1–3 nm	Covalently functionalized (folic acid and Alexa Fluor 488)	HepG2	Selective toxicity by coupling near infrared light and interaction of folic acid‐nanotube and folate receptor‐expressing cancer cells	Folate receptor‐mediated endocytosis	Cytoplasm	41
SWCNT and MWCNT	SWCNT (50–200 nm) and MWCNT (500–2000 nm)	SWCNT (1–3 nm) and MWCNT (10–30 nm)	Covalently functionalized (oxidized) + chitosan, Alexa Fluor 488 and folic acid	HepG2	Little toxic up to 0.01 mg/mL (as shown in Supporting information of reference 38)	Both of endocytosis and piercing/diffusion happened. Addition of folic acid on MWCNTs enabled them to enter the cells via receptor‐mediated endocytosis	Size‐dependent subcellular localization	38
SWCNT	300–1000 nm	1–5 nm	Non‐covalent binding with PEG. EGF or folic acid was conjugated with PEG	SKOV‐3 and OVCA 433	ND	Intact but not fragmented PEG blocked non‐specific uptake of nanotubes; conjugation of EGF or folic acid promoted specific cellular uptake	ND	65
MWCNT (bamboo structure)	3000– 10 000 nm	50–300 nm	Non‐functionalized and suspended in culture medium	Human epidermal keratinocytes	ND	Lectin receptor‐mediated pathway may be involved in cellular uptake	ND	66
SWCNT	500–2000 nm	1.4 nm (10–40 nm after functionalization)	Non‐covalent binding with streptavidin‐labeled quantum dot and biotinylated anti CD3 antibody	Jurkat	ND	Endocytosis dependent on the interaction of CD3 antigen and its antibody	Endocytic vesicles and not found in a nucleus	67
SWCNT	110 nm	10 nm	Covalently functionalized (oxidized, EGF, Qdot and/or cisplatin)	HN12, HN13, SAA and NIH3T3	Selective toxicity dependent on EGFR expression	EGF and EGF receptor‐dependent endocytosis	Perinuclear region	49
SWCNT	<500 nm	1–2 nm	Covalently functionalized (oxidized) + acridine orange	HeLa	Little toxicity	Clathrin‐mediated endocytosis	Lysosomes	68
SWCNT	Short (50–200 nm) or long (200–2000 nm)	Small (1–5 nm) or large (3–15 nm)	Covalently functionalized (oxidized) + streptavidin or albumin, or non‐covalent binding with DNA	HeLa and HL60	ND	Clathrin‐mediated endocytosis	ND	52
MWCNT	100–10 000 nm	10–15 nm	Covalently functionalized (oxidized) + human serum albumin labeled with FITC	HepG2 and CRL‐4020	Selective toxicity by coupling near infrared light and interaction of albumin and gp60‐expressing hepatic cancer cells	Endocysosis; co‐localization with Gp60 (albumin‐binding protein) and caveolin‐1	Perinuclear localization (as shown in figure)	46
MWCNT	10 000–30 000 nm	20–30 nm	With or without covalent functionalization. Nanotubes were dispersed in culture medium containing BSA and DPPC	BEAS‐2B	Induction of IL‐1β	ND (Little uptake of carboxylated nanotubes; nanotubes coated with BSA and DPPC were taken up by the cells whereas non‐coated ones were not)	Vesicles	43
MWCNT	10 000–40 000 nm	50–200 nm (DNA surrounding nanotubes)	Non‐covalent binding with DNA separately labeled with Cy3 or folate	KB	No significant cell death observed up to 4 days incubation	ND (Possibly via endocytosis because folate‐folate receptor interaction facilitates nanotube uptake)	ND	69
SWCNT	100–400 nm	0.7–32 nm	Non‐covalent binding with 29‐amino acid peptide, folding into an amphiphilic alpha‐helix and suspended in culture medium	HeLa	Little toxicity	Time‐ and temperature‐dependent fasion (possibly endocytosis)	Cytoplasm	70
SWCNT	100–400 nm	5–20 nm	Non‐functionalized and suspended in culture medium	HeLa	Little toxicity	Temperature‐dependent mechanism (possibly endocytosis)	Cytoplasmic vacuoles	71
SWCNT	130–660 nm	ND	Non‐covalent binding with DNA	NIH3T3	ND	Size‐dependent uptake of nanotubes; hypothesized as receptor‐mediated endocytosis	ND	37
MWCNT	7000–10 000 nm	80–130 nm	Non‐functionalized and suspended in PBS containing 0.1% gelatin, 0.1% CMC or DPPC	MESO‐1 and BEAS‐2B	Toxicity dependent on dispersant. IL‐6 and IL‐8 production upon nanotube internalization	Nanotubes dispersed in gelatin or DPPC were internalized possibly via endocytosis but nanotubes in CMC were not	ND	45
SWCNT	Several microns	0.7–2.1 nm	Covalently functionalized (amino group) + plasmid DNA	MCF‐7	Little toxicity	Hypothesized as endocytosis without experimental evidence	ND	72
Piercing and diffusion
SWCNT	300–1000 nm	~1 nm	Covalently functionalized (amino group + peptide of alpha subunit of Gs protein)	3T6 and 3T3	Little toxic up to 10 μM	Piercing and diffusion	Cytoplasm and nucleus	54
SWCNT	400 nm	~1.4 nm	Non‐covalent binding with RNA	MCF‐7	Little toxic up to 0.5 mg/mL	Piercing and diffusion	Cytoplasm and nucleus	73
MWCNT	200 nm	20 nm	Covalently functionalized (amino group) + plasmid DNA	HeLa and CHO	Little toxic up to 1.2 mg/mL	Piercing and diffusion	Cytoplasm and nucleus	55
SWCNT	ND (commercial)	ND (commercial)	Covalently functionalized (amino group or FITC)	Primary murine immune cells: B cells and T cells	Little toxicity	Piercing and diffusion	Cytoplasm	53
SWCNT and MWCNT	SWCNT (300–1000 nm) and MWCNT (500–2000 nm)	SWCNT (1 nm) and MWCNT (20–30 nm)	Covalently functionalized (amino group, acetamido group, FITC, methotrexate and/or amphotericin B)	3T6, 3T3, HeLa, Jurkat, keratinocytes, A549, CHO, HEK293 and MOD‐K	ND	Piercing and diffusion, irrespective of functional groups and cell types	Perinuclear region	56
SWCNT	Length‐fractionated samples (84–411 nm)	ND (commercial)	Non‐covalent binding with DNA	A549, MC3T3‐E1, A10 and IMR90	Toxicity dependent on length: nanotubes shorter than 189 ± 17 nm are toxic	SWCNTs shorter than 189 ± 17 nm enter cells by piercing and diffusion	Cytoplasm	74
MWCNT	Sample 1 (~2000 nm) and sample 2 (~300 nm)	Sample 1 (20–40 nm) and sample 2 (>35–40 nm)	Non‐covalent binding with Pluronic F127 and suspended in cell culture medium	HN9.10e (immortalized mouse hippocampal cells) and PC12	ND	Piercing and diffusion, judging from the fact that sodium azide and low temperature did not hinder cellular uptake	Cytoplasmic vacuoles	75
MWCNT	~1000 nm	20–30 nm	Covalently functionalized (oxidized or amino group) + FITC‐BSA	HEK293	ND	Nanotube clusters were taken up by cells via endocytosis and individual ones via piercing and diffusion. Intracellular nanotubes were relatively short	Cytoplasm, vesicles and nucleus	76
MWCNT	50–500 nm	20–30 nm	Covalently functionalized (oxidized followed by introduction of amino groups) and labeled with FITC via π stacking	Catharanthus roseus cell (plant cell)	Little toxicity	Piercing and diffusion (the average lengths of MWCNTs in cells were around 100 nm)	Cytoplasm, vacuole, plastids and nucleus	77
SWCNT and MWCNT	SWCNT (~100 nm) and MWCNT (1000–2000 nm)	SWCNT (4–5 nm in bundle) and MWCNT (20–30 nm)	SWCNT: Covalently functionalized (oxidized) and labeled with FITC via π stacking	Catharanthus roseus cell (plant cell)	Little toxicity	Piercing and diffusion into cytoplasm and active transport into vacuole	Cytoplasm and nucleus	78
MWCNT	1000–2000 nm	10–30 nm	ND	A549	ND	Piercing and diffusion (no negation of endocytic pathway)	Cytoplasm and vesicles. Not found in a nucleus	79
SWCNT and MWCNT	SWCNT (50–500 nm) and MWCNT (50–500 nm)	SWCNT (0.5–1.5 nm) and MWCNT (10–30 nm)	Non‐covalent binding with phospholipid polyethylene glycol amine	OVCAR8	Little toxicity	Hypothesized as piercing and diffusion without experimental evidence	Cytoplasm	80
MWCNT	Largely up to 10 000 nm	Tangled (12 nm), thin (50 nm) and thick (150 nm)	Non‐functionalized and suspended in saline containing 0.5% albumin	Human peritoneal mesothelial cells, MeT5A and MDCKII	Diameter‐ and rigidity‐dependent toxicity and carcinogenicity: thin and rigid nanotubes were the most toxic and carcinogenic	Piercing and diffusion in a diameter‐ and rigidity‐dependent manner	Cytoplasm and nucleus	17
Entrance mechanism not specified or no evidence of nanotube entry
SWCNT	ND (commercial), but highly aggregated	0.8–1.2 nm	Non‐functionalized and suspended in culture medium	A549	Low acute toxicity (0.015–0.8 mg/mL)	No evidence of SWCNT internalization but ultrastructural changes in the cell morphology were observed	No cellular internalization	81
MWCNT	100–13 000 nm	12 nm	Non‐functionalized and suspended in dipalmitoyl lecithin, PBS or ethanol	A549 and MeT5A	Decrease in metabolic activity without apoptosis	Small indivicdual nanotubes could be internalized but not observed with light or transmission electron microscopy	No cellular internalization	82
MWCNT (bamboo structure)	ND (50 000 nm when synthesized)	100 nm (bamboo structure made the thip thinner)	Non‐functionalized and suspended in culture medium	Human epidermal keratinocytes	Slightly toxic with an induction of IL‐8 release	ND (most nanotubes inside cells are <1 μm in length as shown in the figures with a maximal length of 3.6 μm)	Cytoplasmic vacuoles	83
MWCNT	ND	ND	Covalently functionalized (methotrexate and FITC)	Jurkat	ND	ND (independent from the type of covalently‐conjugated functional groups)	Perinuclear region	84
MWCNT	10 000–20 000 nm	100–150 nm	Non‐functionalized and suspended in PBS containing 0.1% gelatin	BEAS‐2B	LDH release with IL‐6 and IL‐8 secretion	ND	Perinuclear localization	85
MWCNT	300–1000 nm	1 nm	Covalently functionalized (amino group)	A549 and CHO	Little toxicity	ND	Perinuclear localization	86
MWCNT	500–2000 nm	20–30 nm	Covalently functionalized (amino group, FITC and/or carboxy group)	A549	ND (Possibly little toxicity)	Both of piercing/diffusion and endocytotic pathway; predominant piercing in the precense of endocytosis inhibitors	Perinuclear localization, cytoplasm and vesicles	87
MWCNT	ND (500–2000 nm in ref.)	20–30 nm	Covalently functionalized (dendron: alkylated amino groups) + siRNA	HeLa and A549	Little toxic up to 0.08 mg/mL	ND	Perinuclear localization and cytoplasm	88
MWCNT	Short (100–3500 nm) or long (100–12 000 nm)	10–160 nm	Non‐functionalized and suspended in water containing arabic gum	A549	Toxic and induces cell death	ND (smallest nanotubes with a maximal length of 3 μm were internalized: 500 nm in length and 50 nm in diameter as shown in the figure)	Cytoplasm and vesicles; not found in a nucleus	89
MWCNT	ND (commercial)	ND (commercial)	Covalently functionalized (anti‐MUC‐1 aptamer was conjugated following oxidation)	MCF‐7 and Calu‐6	Little toxicity	ND (nanotubes were taken up by cells irrespective of MUC‐1 expression, indicating that the translocation was receptor‐independent)	Cytoplasm and nucleus (as shown in the figure)	90
MWCNT	900 nm	ND (<30 nm, judging from the thickness of TEM section)	Non‐covalent binding with Pluronic F127	PC3 and RENCA (a murine renal cancer cell line)	Selective toxicity by coupling near infrared light	ND	Cytoplasmic vesicle and nucleus	91
MWCNT	ND (commercial)	ND (commercial)	Covalently functionalized (oxidized)	HepG2	ND	ND	Cytoplasm	92
SWCNT	200–400 nm	3.8–7.8 nm in ref.	Non‐covalent binding with PEG; conjugated with CpG	GL261	Little toxicity	ND	Cytoplasm	93
SWCNT and MWCNT	SWCNT (5000–30 000 nm) and MWCNT (10 000–30 000 nm)	SWCNT (1–2 nm) and MWCNT (20–30 nm)	Non‐functionalized and suspended in PBS containing Tween 80	A3, MSTO‐211H and HaCaT	SWCNTs were more toxic than MWCNTs	ND (SWCNT penetrating lymphocyte cell membrane is shown in the figure)	ND	94
MWCNT	1000–13 500 nm	67 nm	Non‐covalent binding with Pluronic F68	BEAS‐2B and CHO‐K1	Toxicity dependent on cell density; various cytokine production	ND (No differentiation between engulfed or membrane‐bound MWCNT)	ND	57
MWCNT	300 nm	30–40 nm	Covalently functionalized (polyamidoamine) + DNA	HeLa and COS‐7	Low cytotoxicity at low concentration; positive charge of PAA‐g‐MWCNTs may contribute to the cytotoxicity	ND (Labeled DNA was localized in perinuclear region)	ND	95

BSA, bovine serum albumin; CMC, carboxymethyl cellulose; DPPC, dipalmytoilphosphatidylcholine; EGF, epidermal growth factor; EGFR, epidermal growth factor receptor; FITC, fluorescein isothiocyanate; IL, interleukin; in ref, described in the reference(s); LDH, lactate dehydrogenase; MWCNT, multi‐walled carbon nanotubes; ND, not described; PBS, phophate‐buffered saline; PEG, polyethylene glycol; siRNA, small interfering RNA; SWCNT, single‐walled carbon nanotube.

Size and Surface Characteristics of Carbon Nanotubes Determine Uptake Pathways

Size‐dependent uptake of carbon nanotubes into cells has been well studied.17, 37, 38 Kang et al.38 showed that SWCNTs, but not MWCNTs, all of which were oxidized and non‐covalently bound with chitosan, mainly entered human HepG2 hepatocellular carcinoma cells via endocytosis. This result indicates that large carbon nanotubes cannot be taken up by cells through endocytosis due to size limitations. However, it is notable that the conjugation of folic acid on MWCNTs enabled them to enter cells, indicating that specific, receptor‐mediated endocytosis allows large fibers to be internalized.38, 39, 40, 41, 42 We recently showed that MWCNTs coated with albumin on their surface were not endocytosed by two different human mesothelial cell lines.17 Instead, we found that these MWCNTs directly pierced through the plasma and nuclear membranes in a diameter‐ and rigidity‐dependent manner, suggesting that large MWCNTs can still enter cells through direct piercing rather than endocytosis. On the other hand, Wang et al.43 showed that MWCNTs coated with both albumin and phospholipid (DPPC, dipalmitoyl‐phosphatidylcholine) were successfully internalized by bronchial epithelial BEAS‐2B cells. Moreover, Haniu et al.44, 45 showed that MWCNTs bound to gelatin or a phospholipid (DPPC) were taken up by mesothelioma MESO‐1 cells, bronchial epithelial BEAS‐2B cells and neuroblastoma IMR‐32 cells. Taken together, current data indicate that MWCNTs, which are comparatively larger than endosomes, may not be endocytosed, but the presence of specific molecules, such as folic acid and DPPC, may help cells take up MWCNTs. Albumin did not facilitate endocytosis in our previous study, but previous reports indicate that albumin may help endocytosis if the target cells express high levels of an albumin receptor, such as gp60.46 Similarly, Jin et al.37 showed that among small SWCNTs, the efficiency of nanotube uptake is also size‐dependent. They showed that SWCNTs wrapped by DNA were most efficiently taken up by NIH3T3 cells when the nanotube length was ~300 nm.

Regarding the size limitation on membrane piercing, it is of note that Kusumi and colleagues reported a membrane skeleton fence model.47, 48 This model explains that an actin mesh network exists just beneath the cell membrane to form compartments whose size are dependent on cell type (side length is 30–230 nm). Based on this model, we can expect that very thick nanotubes (>230 nm) that evade endocytosis due to their surface property cannot directly pierce the membrane.

Surface characteristics are also critical in carbon nanotube uptake. For instance, oxidized nanotubes are more likely to enter cells via endocytosis with the help of specific ligands42, 46, 49, 50, 51, 52 because the oxidation of nanotubes introduces carboxyl groups, which provides a negative charge to their surface, resulting in repulsion from the plasma membrane unless the carbon nanotube has ligands that are recognized with a high affinity by the cell. Many ligands (folic acid,42 albumin46 and epidermal growth factor49) have been reported to facilitate uptake by specific cell types. Alternatively, a carbon nanotube with amino groups presents a positive charge on its surface and can enter cells via energy‐independent piercing and diffusion.53, 54, 55 On the other hand, Kostarelos et al.56 showed that MWCNTs with functional groups entered various types of cells in a functional group‐independent manner via membrane piercing and diffusion. In our own studies we have found that pristine MWCNTs with albumin on their surface pierced the cell membrane in a diameter‐dependent manner.17 These results indicate that a positive charge on the surface of carbon nanotubes might not be necessary for the carbon nanotubes to pierce through the cell membrane. However, the detailed mechanism remains to be elucidated.

Carbon Nanotubes and Asbestos Fibers Activate Macrophages

Macrophages are a type of immune cell that function in the recognition and removal of pathogens and exogenous materials. Macrophages sometimes fail to achieve this goal when the pathogens are too large and/or resistant to biodegradation. Certain nanotubes and asbestos fibers have these features, leading to persistent macrophage activation and enhanced chronic inflammation.5, 6

It has been reported that macrophages recognize asbestos fibers and carbon nanotubes via the class A scavenger receptor, MARCO (macrophage receptor with collagenous structure).57, 58, 59, 60 Moreover, the uptake of these two types of materials activates the NLRP3 inflammasome.61, 62 Therefore, asbestos fibers and carbon nanotubes appear to activate macrophages through similar mechanisms. It is noteworthy that the carbon nanotubes used in these experiments were non‐functionalized.

Concluding Remarks

A number of sequential events are involved in fibrous material (i.e. carbon nanotubes and asbestos fibers) toxicology: translocation of materials, biopersistence of materials, mesothelial/epithelial cell injury induced by materials and macrophage activation triggered by materials. Among these, we focused on how asbestos fibers and carbon nanotubes enter non‐phagocytic cells, which is important in mesothelial/epithelial cell injury.96, 97 Asbestos fibers are endocytosed by these non‐phagocytic cells and carbon nanotubes behave variably according to their functionalization and size (Fig. 2). Based on current evidence, a positive charge on the surface and a thin diameter appear to be two important factors in facilitating the membrane piercing ability of carbon nanotubes. The presence of specific ligands on the carbon nanotube surface would induce ligand‐mediated endocytosis, which allows large‐sized nanotubes to enter non‐phagocytic cells. Further investigation regarding the mechanism of entry of carbon nanotubes and asbestos fibers into non‐phagocytic cells is warranted to establish the distinct toxicological paradigms of carbon nanotubes and asbestos fibers.

Figure 2.

A schematic of the fiber entry mechanism into a non‐phagocytic cell. Asbestos fibers are endocytosed by the cells through phagocytosis‐like mechanisms. Mechanisms of carbon nanotube uptake depend on the characteristics of each fiber. Nanotubes with thin diameter (~50 nm in Fig. 1), high rigidity and positive surface charge are likely to pierce the plasma and nuclear membrane whereas nanotubes with shorter length (<1 μm) and specific ligands on their surface with modification would be endocytosed by the cells.

Disclosure Statement

The authors have no conflict of interest.