Alternative Estimation of Human Exposure of Single-Walled Carbon Nanotubes Using Three-Dimensional Tissue-Engineered Human Lung

Abstract

Recent discoveries of various forms of carbon nanostructure have stimulated research on their applications and hold promise for applications in medicine and other related engineering areas. Although carbon nanotubes (CNTs) are already being produced on a massive scale, few studies have been performed which test the potential harmful effects of this new technology. The authors used a three-dimensional in vitro model of the human airway using a coculture of normal human bronchial epithelial cells and normal human fibroblasts for the health risk assessment of CNTs on the human respiratory systems. The authors used aqueous single-walled carbon nanotube (SWCNT) solution. The average length and diameter of nanotube ropes were about 500 nm and less than 10 nm, respectively. The authors measured the production of nitric oxide (NO) as an inflammatory marker and mitochondrial activity using MTT (3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) assay as a cytotoxic response of the cell layers following exposure of different concentration of aqueous SWCNT solution. The results indicated that NO production was dramatically increased and cell viability was decreased following exposure of different concentrations of SWCNTs. Transepithelial electrical resistance (TER) across the coculture layers was measured to observe the changes in airway physiological function following exposure of different concentrations of SWCNTs. TER value was dramatically decreased following exposure of 20% SWCNT (8 μg/ml). In this study, the authors presented viable alternatives to in vivo tests to evaluate the toxicity of engineered SWCNTs. Cytotoxic/inflammatory responses and barrier function of the human lung layers following exposure of SWCNTs were observed using in vitro coculture system of airway.

The discovery of carbon nanotubes (CNTs) in 1991 initiated a great number of interests in various scientific and technological fields. Owing to their phenomenal, mechanical, electrical, and thermal properties, many potential applications have been proposed for CNTs, including conductive and high-strength composites, energy storage devices, sensors and actuators, field emission displays, nano-scale semiconductor devices, probes, and interconnects (Iijima 1991). The CNT manufacturing processes and the post processes that transform the raw material into value-added products entail substantial handling of CNTs in powder form and expose the workers to the risk of inhalation or skin contact. Recent studies for nanomaterials indicate (1) CNTs and fullerenes have produced toxic effects on biological systems (Chin et al. 2007; Dumortier et al. 2006; Helland et al. 2007; Lam et al. 2006; Yang et al. 2006); (2) evidence that nanoparticles can translocate to bloodstream (Rothen-Rutishauser et al. 2007; Shimada et al. 2006); and (3) evidence that nanoparticles can cross blood-brain barrier (Kim et al. 2007). However, studies are still preliminary, as the current in vivo and in vitro response data are difficult to extrapolate, and coating (or surface treatment) influences how particles interact with biological systems. Intensive studies on the toxicity of CNTs have shown that exposure to CNTs results in pulmonary inflammation (Chou et al. 2008; Shvedova et al. 2005; Warheit et al. 2004; Muller et al. 2005; Mitchell et al. 2007; Lam et al. 2004; Li et al. 2007). The inflammatory lung reactions (alveolitis) are a source of genetic lesions, which could eventually lead to the development of lung cancer (Chou et al. 2008). In vivo studies performed using guinea pigs and rats showed the appearance of multifocal granulomas, resulting in inflammatory reactions of the terminal and respiratory bronchials. Mild fibrosis in the alveolar septa was also observed (Helland et al. 2007).

Ken Donaldson and his colleagues described three properties of CNTs associated with pathogenicity in particles. They are (1) nanoparticles showing more toxicity than larger sized particles, (2) fiber-shaped particles behaving like asbestos and other pathogenic fibers that have toxicity associated with their needle-like shape, and (3) biologically biopersistent. They also pointed out that CNTs are possibly one of the least biodegradable man-made materials ever devised (Donaldson et al. 2006). Also concerns over the increased emissions of CNTs into the environmental compartments (air, water, and soil) mainly due to improper disposal of CNTs were raised (Helland et al. 2007).

Experimental access to the airways is, in general, very difficult. Postmortem analysis of the smaller airways and direct bronchial biopsy of the upper airways (approximately generations 2 to 5) have provided a wealth of information. Unfortunately, supply of these tissues, and experimental control of the system, are limited. Monolayer culture of the key individual cell types has also provided abundant fundamental information on the response of these cells (e.g., epithelial, fibroblast, and smooth muscle) to external perturbations. These systems are limited by the absence of cell-cell interactions, which are present in vivo. More recently, a contracting fibroblast-embedded collagen gel has provided a more dynamic model to simulate human airway (Adachi et al. 1998; Fredriksson et al. 2002; Infeld, Brennan, and Davis 1992; Agarwal et al. 2001; Mio et al. 1998; Umino et al. 2000; Zhang et al. 1999). Fibroblasts will naturally contract the extracellular matrix (ECM) to close a wound, and, when placed in a collagen gel, respond in a similar fashion and contract the gel. Another in vitro model of airway involves culturing epithelial cells as monolayer on a membrane and fibroblasts as a monolayer a fixed distance away separated by culture medium (Morishima et al. 2001; Swartz et al. 2001). This model is attractive as it isolates soluble mediators that participate in epithelial-fibroblast communication, but lacks the normal ECM and anatomical dimensions. Our model of the airway in this study presented several important features: (1) It maintains the normal anatomical arrangement (orientation and dimensions) of epithelial and fibroblast cells. (2) The fibroblast is embedded in collagen I, yet remains anchored. (3) A thin (10-μm) porous polyester membrane separates the epithelial and fibroblast cell, allowing not only communication between the epithelium and fibroblast, but also clean access to investigate cell-specific protein expression, following exposure to external perturbation.

In this study, we presented viable alternatives to in vivo tests to evaluate the toxicity of engineered single-walled carbon nanotubes (SWCNTs). Cytotoxic/inflammatory responses and barrier function of the human lung layers following exposure of CNTs were observed using in vitro coculture system of airway. Further studies are required to get the transport properties of CNTs across the cells and specific gene expression pattern either in monolayer or in coculture. In addition to inflammatory and cytotoxic responses of cell layers, the changes in physiological functions, such as mucin secretion, tight junction formation, and cilia formation, need to be measured following exposure of different concentrations and structures of CNTs.

MATERIALS AND METHODS

Characterization of Carbon Nanotubes

The high-purity single-walled carbon nanotubes (SWCNTs) used in this study were produced by Carbon Nanotechnologies Inc., using the HiPCO process. The residual metal content was 3% to 12% by weight, and the individual nanotubes were 0.8 to 1.2 nm in diameter and 100 to 1000 nm in length (manufacturer provided data). The SWCNTs were added to distilled water, and the mixture was sonicated for 60 min using Sonicator 3000 manufactured by Misonix (Farmingdale, NY). A nonionic octylphenol ethoxylate surfactant, Triton X-100, was added to the dilute aqueous SWCNT solution to facilitate the separation of nanotube ropes into smaller ropes or individual tubes.

The degree of SWCNT dispersion in the aqueous solution was evaluated using an atomic force microscope (AFM) MultiMode II developed by Veeco Digital Instruments Group (Woodbury, NY). Several drops of nanotube suspension were applied on to a silicon substrate and were allowed to dry in open air, leaving nanotube agglomerates. The substrate was observed under the AFM, and the nanotube rope sizes were measured at various locations.

Preparation of Human Bronchial Epithelial Cells and Human Lung Fibroblasts for Three-Dimensional Coculture

The human airway lining cell layers play a role as a barrier to external stimuli. The airway consists of several cell layers, such as epithelial cells, fibroblasts, and smooth muscle cells, including several inflammatory cells (e.g., macrophages, neutrophiles, mast cells, etc.), as shown in Figure 1. Physiological response to external perturbation is induced by each cell layers and/or the interaction between cell layers. Our in vitro coculture system was established based on the interaction between two cell layers such as epithelial cells and fibroblasts shown in Figure 2.

Fibroblast-embedded collagen I gels were prepared using rat tail tendon collagen (RTTC; Collaborative, Bedford, MA). Normal human lung fibroblasts (NHLFs) were harvested upon reaching 75% to 80% confluency, and added (seeding density of 5 × 10 fibroblasts/ml of final volume) to an iced mixture of collagen (final concentration 2.0 mg/ml), 5× concentrated Dulbecco’s modified Eagle’s medium (DMEM), and 10× reconstitution buffer comprised of NaHCO₃, HEPES buffer (Gibco, Grand Island, NY), and NaOH. Aliquots of the mixture were pipetted onto the underside of a Transwell (Costar, Cambridge, MA) polyester membrane (0.4-μm pore). The collagen mixture is then allowed to “gel” (noncovalent cross-link) at 37°C in 5% CO₂ for 10 to 15 min (Figure 2 A ). Harvested primary human bronchial epithelial (HBE) cells (passages 2 to 3) were then seeded (1.5 × 10 cells/cm²) directly on top of the polyester membrane. The entire tissue was submerged in medium for 5 days and the epithelium was allowed to attach and became confluent (Figure 2B ). For the first 48 h, the medium was basal epithelial growth medium (BEGM; Clonetics) and a low retinoic acid concentration. For days 3 to 5 (and days 6 to 21), the medium was a 50:25:25 mixture of BEGM:DMEM:Hams F12, with a high retinoic acid concentration. At day 6, an air-liquid interface was established (medium maintains a high retinoic acid concentration) and the epithelium was allowed to differentiate for approximately 2 weeks, at which time it was ready for experimentation (Figure 2C ).

Measurement of Transepithelial Electrical Resistance (TER)

Human bronchial epithelial cells were grown at the interface of air and liquid. Culture medium was provided from the bottom through the porous membrane. TER of human bronchial epithelial cell with fibroblasts-embedded collagen layers cultured in Transwell was monitored using a portable Voltohm-meter (Millipore, Bedford, MA) attached to a dual “chopstick” or transcellular resistance measurement chamber (Millipore). Different concentrations of CNTs were exposed to the coculture layers for 6 h. Each of the two electrode systems contained Ag/AgCl electrode for measuring voltage and a concentric spiral of silver wire for passing current across the epithelium. Electric current could then be passed across the epithelium to measure TER (Ω·cm²). It is perceived that TER values higher than the background fluid resistance indicate a confluent airway epithelium with tight junctions. TER was monitored to identify the perturbation in the normal physiology and and permeability of human bronchial epithelial cells (Rejman et al. 2007).

Cytotoxicity

The MTT (3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetra-zolium bromide, a tetrazole) assay (Sigma) was used to evaluate the changes in cellular metabolic (mitochondrial) activity of cells as a cytotoxic response. Cells were exposed to varying concentrations of SWCNTs (Table 1). After 48 h, 150 μl of MTT (5 mg/ml) was added to each well and incubated for 4 h. Afterward, 850 μl of the MTT solubilization solution (10% Triton X-100 in 0.1 N HCl in anhydrous isopropanol) was added to each well. The resulting formazan crystals was solubilized in acidic isopropanol and quantified by measuring absorbance at 570 nm. Data were calibrated to the appropriate calibration curve as stated in Sigma protocols.

Inflammatory Response

Nitric oxide (NO) is produced by many cells in the body. Under normal (basal) conditions, NO is continually being produced by cNOS (constitutive nitric oxide synthase). However, during inflammation, the amount of NO produced by iNOS may be a 1000-fold greater than that produced by cNOS. NO production was measured to identify the level of inflammation (Clancy, Amin, and Abramson 1998). All media samples were analyzed using Griess Reagent system (Promega, WI) to detect the level of nitrite ( ${\text{NO}}_2^{-}$ ), one of the two stable oxidized products of NO in a liquid phase (Bredt and Snyder 1994).

Statistical Analysis

Statistical analyses were carried out using one-way analysis of variance (ANOVA) followed by Dunnett’s multiple comparison test (Cheung and Holland 1991) to determine where significance exists (p < .05).

RESULTS

Characterization of Carbon Nanotubes

Figure 3 shows 100 ml of aqueous SWCNT suspension 12 h after sonication. Dark, uniform solution suggests the nanotubes are well dispersed in the medium. The surfactant-aided nanotube suspension remained stable for at least 2 months. The AFM image of SWCNTs dried on a silicon substrate is shown in Figure 3. The average length and diameter of nanotube ropes were about 500 nm and less than 10 nm, respectively. A more detailed study on the distributions of nanotube dimensions using image analysis is discussed in Wang et al. (2007).

Effect of CNTs on Physiological Function of Airway Epithelial Cells

Different concentrations of SWCNTs were exposed to the coculture layers for 6 h. The TERs of the controls (5% and 20% of Triton X-100 and 0% of SWCNT) were stable around 500 Ω·cm² (resistance of epithelial-free tissue was subtracted) for 48 h. Ten percent to 20% of SWCNTs rapidly compromised the barrier function of the epithelium and the TER decreased to 120 Ω·cm². After removing SWCNTs, the TERs completely recovered to the control level (Figure 4).

Inflammatory and Cytotoxic Responses

NO production following exposure of SWCNTs to epithelial cells was dramatically increased as the concentration of SWC-NTs increased (Figure 5A ). At higher concentrations of SWC-NTs, cells showed cytotoxic response and parts of cell layers were detached (data not shown). Each NO production was normalized by total proteins. Fibroblasts showed inflammatory response, slightly increasing the level of NO following exposure of SWCNTs (Figure 6A ). Cellular metabolic activity was observed following exposure of different concentrations of SWCNTs to both cell layers. MTT activity was decreased as concentration of SWCNTs increased, especially for epithelial cells (Figures 5B and 6B).

DISCUSSION

In this study, we developed a viable alternative to in vivo tests to evaluate the toxicity of engineered CNTs. Experimental access to the airways involved in health risk of CNTs is, in general, very difficult. Monolayer culture of the key individual cell types has also provided abundant fundamental information on the response of these cells (primarily epithelial, fibroblast, and smooth muscle) to perturbations. These systems are limited by the absence of cell-cell interactions, which are present in vivo. We used a new coculture technique that offers several distinct advantages over earlier models (Adachi et al. 1998; Fredriksson et al. 2002; Infeld, Brennan, and Davis 1992; Agarwal et al. 2001; Mio et al. 1998; Umino et al. 2000; Zhang et al. 1999; Nimni 1983; Arora and McCulloch 1994; Singer et al. 1984; Hunt et al. 1994). The airway epithelial cells were cultured as a monolayer over a thin (10-μm porous polyester membrane. A thin lung fibroblast-embedded collagen layer was attached to the opposing side of membrane. In this fashion, the epithelial cells and fibroblast cells maintained the normal anatomical arrangement, but the polyester membrane allowed easy separation of the cell types for cell-specific gene expression and proteomics analysis. In addition, the air-liquid interface was more easily controlled, and the collagen gel was anchored thus maintaining tension in the matrix that more closely resembled the in situ condition. Figure 2 depicts schematically how the tissue was generated. Our coculture model in this study showed the expression of three important phenotypic markers such as mucin (mucous production), F-actin (tight junction), and tubulin (cilia) (Data not shown).

Although human lung carcinoma epithelial cell line (A549) provides a valuable information regarding in vitro cytotoxicity of SWCNTs (Davoren et al. 2007), these transformed cell lines are typically less resistant to toxic effects than other cells derived from normal tissues, and also do not fully exhibit normal phenotypic markers. To test how SWCNTs affect the barrier function of lung epithelial cell layers, we used normal human bronchial epithelial cells, which can form tight junction, to observe the perturbation of TER following exposures of SWCNTs. TER of our coculture layers was measured to observe the changes in airway physiological function following exposure of different concentrations of SWCNTs. The TER value was dramatically decreased following the exposure of 20% CNT. This means high concentration of CNTs debilitated the barrier function of airway epithelial cell layers. We did not observe the translocation of SWCNTs through the individual cells or cell layers microscopically. However, our TER data (especially, the decrease in TER at high concentration of SWCNTs) indirectly told that SWCNTs can possibly penetrate thorough the cell layers. Translocation through the individual cell, accumulation, and accumulation after translocation of SWCNTs need to be observed microscopically using SEM (scanning electron microscopy) in the future. There was no significant morphological change observed under phase contrast microscopy in this study.

We also measured the production of nitric oxide (NO) as an inflammatory marker and MTT activity for cytotoxic response of the cell layers following exposure of different concentrations of SWCNTs. The airway epithelial cell showed inflammatory response as they increased NO production, and also showed cytotoxic response, whereas fibroblasts showed mild inflammatory and cytotoxic response. Nonionic detergent (Triton X-100) used to avoid the aggregation of SWCNTs did not show a significant cellular toxicity and inflammatory response. In order to distinguish between SWCNT toxicity and general particle toxicity, quartz particles was tested in our initial experiments. Quartz particles (as a positive control) with a particle size distribution of 0.35 to 3.50 mm at same concentration range did not show any significant inflammatory and cytotoxic response (data not shown) in this study (Yacobi et al. 2007; Davoren et al. 2007). Further studies about the effects of shape, size, exposure time, and aggregation of CNTs on airway function and cellular toxicity need to be investigated. During 2-month period of our experimentation, CNT aggregation was not detected with phase contrast microscope. However, SWCNT aggregation was observed 2 months after we made the initial solution of SWC-NTs with Triton X-100 (Figure 7). In this study, the effect of SWCNT aggregation on airway physiological function and cellular toxicity was not investigated. CNT aggregation or extent of aggregation may generate an adverse effect on cell function and viability (Soto, Garza, and Murr 2007). In addition to the health risk evaluation of SWCNTs alone, effects of sequential exposure to SWCNTs and other pollutants were also investigated using mice. Exposure to CNTs may occur in conjunction with other pathogenic impacts (microbial infection) and trigger enhanced responses (Shvedova et al. 2008). Functional respiratory deficiency and decreased bacterial clearance were found in mice with SWCNTs (Shvedova et al. 2007). CNT-induced cytotoxic/inflammatory responses were attenuated in mice following exposure to both CNTs and low level of ozone (Han et al. 2008).

In this study, we presented viable alternatives to in vivo tests to evaluate the toxicity of engineered SWCNTs. Our in vitro coculture configuration separated the epithelial and fibroblast cell allowing communication between the epithelium and fibroblast, but also clean access to investigate cell-specific protein expression, following exposure to external perturbation, in the human respiratory system. Using this coculture system, we will be able to address the following important questions: (1) How do two different types of lung cells interact with each other to respond to SWCNT exposure? (2) What is the cellular and molecular mechanisms of cytotoxic response and interaction in the human respiratory system? and (3) How different size and structure of SWCNTs will be translocated and accumulated to alter the mechanisms of cellular response and specific gene expression pattern? Cytotoxic/inflammatory responses and barrier function of the human lung layers following exposure of SWCNTs were observed using in vitro coculture system of airway. Further studies on nanoparticle-related cellular toxicity and functional relations between the size or structure of SWCNTs and the perturbation of cellular or physiological functions are required in the future.