Functionalization Density Dependent Toxicity of ... - ACS Publications

Sep 20, 2012 - was assessed in a murine macrophage RAW 264.7 cell line, a model for liver Kupffer cells. In vitro cytotoxicity of oxidized. MWCNTs was...
0 downloads 0 Views 3MB Size
Download PDF
Article pubs.acs.org/crt

Functionalization Density Dependent Toxicity of Oxidized Multiwalled Carbon Nanotubes in a Murine Macrophage Cell Line Raman Preet Singh, Manasmita Das, Vivek Thakare, and Sanyog Jain* Centre for Pharmaceutical Nanotechnology, Department of Pharmaceutics, National Institute of Pharmaceutical Education and Research (NIPER), SAS Nagar (Mohali) Punjab, India 160062 S Supporting Information *

ABSTRACT: The present study investigates the effect of functionalization density on the toxicity and cellular uptake of oxidized multiwalled carbon nanotubes (f-MWCNTs) in vitro. The toxicity of f-MWCNTs at varying degrees of carboxylation was assessed in a murine macrophage RAW 264.7 cell line, a model for liver Kupffer cells. In vitro cytotoxicity of oxidized MWCNTs was directly proportional to their functionalization density. The increased cytotoxicity was associated with a concurrent increase in the number of apoptotic cells and production of reactive nitrogen species (RNS). In contrast, reactive oxygen species (ROS) generation was the highest in the case of pristine MWCNTs and decreased with increased functionalization density. Quantitative cellular uptake studies indicated that endogenous ROS production was independent of the concentration of CNTs internalized by a specific cell population and was directly proportional to their surface hydrophobicity. Mechanistic studies suggested that cellular uptake of CNTs was critically charge-dependent and mediated through scavenger receptors, albeit the involvement of nonscavenger receptor mechanisms at low CNT concentrations and their saturation at the experimental concentration cannot be ruled out. A mathematical model was established to correlate between the cellular uptake of CNTs with their length and zeta potential. In an attempt to correlate the results of in vitro toxicity experiments with those of the in vivo toxicity in the mouse model, we found that the toxicity trends in vitro and in vivo are rather opposing. The apparent anomaly was explained on the basis of different experimental conditions and doses associated with cells under in vivo and in vitro culture conditions.

1. INTRODUCTION Over the course of the past decade, the field of Nanomedicine has substantially matured, offering potential solutions for improved drug and gene delivery. The increased interest in nanoparticles (NPs) is largely attributed to their small size which accounts for increased NP−cell interactions and subsequently their high accumulation inside the target cells. In some of our earlier reports, we have already demonstrated that anticancer drug loaded polymeric,1,2 protein,3 and multimodal magnetic4−7 NPs exhibit enhanced efficacy compared to that of the free drug due to their highly specific intracellular accumulation. The use of multimodal NPs, in particular, offers exciting new opportunities that have remained unexplored with conventional delivery systems. Multimodal nanoparticles not only facilitate the delivery of therapeutic molecules to their pathological site of interest but also concurrently enable a real-time monitoring of drug delivery via molecular imaging.4,5 Among the broad spectrum of pharmaceutical nanocarriers that are able to enhance the delivery, absorption, and intracellular uptake of a bioactive molecule while protecting it from deactivation, carbon nanotubes (CNTs) stand to be an attractive vehicle for multimodal delivery of therapeutic compounds. The increased © 2012 American Chemical Society

interest in CNTs may be attributed to their high aspect ratio and surface area, ultralight weight, high thermal conductivity, mechanical strength, ease of drug-loading via π−π stacking interaction, near-infrared fluorescence, Raman scattering, photoacoustic effects, and, above all, their natural shape, which facilitates their noninvasive penetration across biological barriers through a nanoneedle-type mechanism. In addition to all these advantages, CNTs offer a tunable surface chemistry, which makes them an efficient and versatile vehicle for covalent conjugation with a diverse array of functional molecules. As for the biofunctionalization of CNTs, acid-oxidized, carboxylated CNTs deserve special mention because their ease of preparation, ready interchangeability with other functional groups, and easily manipulable surface functionality with adjustable functionalization density make them an attractive bioconjugation precursor for drug delivery vehicles and other CNT-based biomedical devices.8,9 Since carboxylated CNTs are used as the base-material for a majority of CNT-based biomedical platforms, a thorough assessment of their toxicity is necessary. In most of the bioconjugation reactions involving Received: May 18, 2012 Published: September 20, 2012 2127

dx.doi.org/10.1021/tx300228d | Chem. Res. Toxicol. 2012, 25, 2127−2137

Chemical Research in Toxicology

Article

2.2. Preparation of Functionalized MWCNTs. 2.2.1. Surface Oxidation of Pristine MWCNTs. For surface oxidation, pristine MWCNTs (50 mg) were dispersed in 20 mL of H2SO4/HNO3 at a ratio of 3:1 (v/v). The dispersion was sonicated for 5 min to debundle the aggregates and get a uniform dispersion. Thereafter, the dispersion was refluxed at 80 °C using a hot plate stirrer at 900 rpm. Oxidation was carried out over a period of 1−8 h to ensure different degrees of functionalization on the surface of MWCNTs. Following oxidation, dispersion in acids was diluted up to 5 times its volume and centrifuged to isolate the functionalized product from the mother liquor. This washing process was continued (3−4 times) until complete removal of the acid. The pellets of functionalized MWCNT (f-CNT) were dispersed in acetone, recentrifuged, and finally dried in a vacuum oven at 50 °C. 2.2.2. Preparation of Poly-L-lysine (PLL)-MWCNTs. Poly-L-lysine (PLL) coated MWCNTs (CNT-PLL) were prepared by simple mixing of MWCNTs (1 mg) with PLL (10 mL; 0.01%) for 12 h. MWCNTPLL was separated by centrifugation (10,000g; 30 min) and washed twice with water to remove excess PLL. 2.2.3. Preparation of PEG-MWCNTs. For the preparation of PEGylated MWCNTs, acid-oxidized MWCNTs (20 mg) were dispersed in tetrahydrofuran (THF, 2 mL) and refluxed with thionyl chloride at 65 °C. Following 24 h of reaction, thionyl chloride was removed in vacuum, and acylated MWCNTs were flooded with 5-fold excess of methoxy-PEG-OH (dissolved in 1 mL of DMSO). The reaction was continued for 12 h, following which the CNTs were isolated by repeated washing with water and acetone and finally desiccated in a vacuum oven. 2.3. Physicochemical Characterizations. The morphology of CNTs during functionalization (oxidation) was followed up using scanning electron microscopy. Zeta potential was measured using the Malvern Zeta Sizer (Nano ZS, Malvern Instrument, US). Surface chemistry of the functionalized CNTs was studied using Fourier transform infrared (FTIR) spectroscopy and thermogravimetric analysis (TGA). FTIR spectra were recorded on Perkin-Elmer systems using KBr pellets and processed using Spectrum Software. TGA was carried out on a Perkin-Elmer System by heating 5 mg of p- and fCNT at the rate of 10 °C/min. The contact angle was recorded by a contact angle goniometer using the static sessile drop method. 2.4. In Vitro Cytotoxicity and Cell Uptake Studies. RAW 264.7 cells (1 × 104 cells/well) were grown in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% v/v fetal calf serum as described earlier.19 Cell suspension (5 × 105 cells/mL; 200 μL) was added in 96-well tissue culture plates and incubated overnight for cell attachment. CNTs were dispersed in culture medium to obtain a stock dispersion of 1 mg/mL. The stock dispersion was bath sonicated for 5 min to aid dispersion. The stock dispersion was serially diluted in culture medium to achieve target CNT concentrations. The culture medium of wells was replaced with 200 μL of culture medium containing appropriate concentrations of CNTs, and toxicity was determined as follows. 2.4.1. Cell Viability Studies. For the evaluation of in vitro cytotoxicity, RAW 264.7 cells (1 × 104 cells/well of 96 well tissue culture plate) were incubated with various concentrations of MWCNTs (0−100 μg/mL) up to 96 h, and cell viability was determined by the MTT assay as described earlier.1 Briefly, 5−6 wells/ concentration were incubated with 0.5 mg/mL MTT in DMEM for 3 h, and formazon was dissolved in 200 μL of DMSO. The optical density (OD) was determined at 550 nm; this OD represented contribution by formazon as well as CNTs (ODMTT+CNT). In a parallel set of experiments, 3−4 wells/concentration were incubated with DMEM alone (without MTT) for 3 h, and cells were dissolved in DMSO. The OD at 550 nm accounted for the contribution by CNTs alone (ODCNT). The mean of all ODCNT values was calculated and is represented as mODCNT. The mODCNT value for each CNT concentration was subtracted from each ODMTT+CNT value using the following formula:

oxidized CNTs, functional bioactives are appended to the nanotubes via hydrolyzable linkages. As depicted by some earlier reports, covalently functionalized CNTs are highly susceptible to defunctionalization in the liver via chemoenzymatic hydrolysis or radical attack and may return to oxidized CNTs.10 Therefore, the toxicity associated with any covalently functionalized CNT platform, in the long term, may be a consequence of their defunctionalization into oxidized CNTs. Consequently, the safety and biocompatibility of oxidized MWCNTs need to be carefully assessed while employing the same for drug delivery and relevant application. Over the past few years, a number of reports have embarked on the toxicity of acid-oxidized, carboxylated MWCNTs both in vitro and in vivo.11−13 As depicted by many of these reports, an alteration in the degree of functionalization may severely transform the cell responses to CNTs and alter their toxicity profile at both the extracellular14 and intracellular levels.15 In an earlier study from our group, we had shown that the toxicity of intravenously injected oxidized multiwalled carbon nanotubes (MWCNTs) in mice critically depended on their functionalization density.16 We observed that acid oxidized, carboxylated CNTs with a higher number of surface carboxyl groups were more hydrophilic, biocompatible, and subsequently less toxic than their pristine as well as oxidized counterparts with lower degree of functionalization. Of note, our previous study was restricted to the evaluation of general organ-level toxicity as well as the determination of various biochemical markers, indicative of hepato-and nephrotoxicity. We, however, did not address whether and how the functionalization density of CNTs influences their toxicity at the cellular level and whether any in vitro−in vivo correlation is possible. Moreover, we have not come across any primary literature that correlates the functionalization density of oxidized CNTs with their toxicity behavior in vitro. We therefore sought to investigate the effect of functionalization density on the toxicity of CNTs in vitro (preferably in some cell line which models the Kupffer cells). Taking RAW 264.7 (mouse leukemic monocyte macrophage cell line) as a model of Kupffer cells,17,18 we tried to apprehend how the surface carboxyl density of CNTs influences their cellular viability, apoptotic activity, and free radical generation profile. Functionalization density dependence of CNTs’ cellular uptake was studied. A quantitative, mathematical model was developed to correlate the length and surface charge of CNTs with the observed cellular uptake. Mechanistic studies were conducted to address the enhanced cellular uptake of negatively charged CNTs. Finally, we tried to correlate the in vitro toxicity results with our previously reported in vivo data and examine whether in vitro models can be used as reliable predictors of CNTs’ toxicity in vivo.

2. MATERIALS AND METHODS 2.1. Chemicals and Cells. Pristine multiwalled CNTs (MWCNTs) were a generous gift from Nanovatec, Pvt. Ltd., US. Dulbecco’s modified Eagle’s medium (DMEM), fetal bovine serum (FBS), and the antibiotic−antimycotic mixture was purchased from PAA, Austria. Apoptosis detection kit (Annexin-V), dialysis membrane, dichlorofluorescein diacetate (H2DCF-DA), MTT, N-1-napthylethylenediamine dihydrochloride, phosphoric acid and sodium nitrite were purchased from Sigma, USA. The TUNEL kit was purchased from Calbiochem, USA. 5-(4, 6-Dichlorotriazinyl) aminofluorescein (DTAF) was purchased from HiMedia, India. All other reagents used were of analytical grade. RAW 264.7 cells were purchased from National Centre for Cell Sciences, Pune, India.

ODadj = ODMTT + CNT − mODCNT 2128

dx.doi.org/10.1021/tx300228d | Chem. Res. Toxicol. 2012, 25, 2127−2137

Chemical Research in Toxicology

Article

where, ODadj is the adjusted OD used for determining cell viability. All ODCNT values were very close, and the standard deviation of mODCNT was 95% cell death. These results suggest that oxidized CNTs are toxic to cells only when incubated for a longer duration (>48 h). Functionalization density dependent reduction in cellular viability is observed only at or above a concentration of 10 μg/mL. Even at or below this concentration limit, cytotoxicity exerted by 0−4 h oxidized

of oxidation were not retained in the liver/spleen and rapidly cleared out from the systematic circulation through the renal excretion route without inducing any obvious nephrotoxicity. Using exactly the same batches of MWCNTs with the same physicochemical properties that had been reported in our earlier publication, we tried to investigate the functionalization density dependence of MWCNTs’ toxicity and cellular uptake in vitro. The results of CNT characterizations (particle characteristics and surface chemistry) are presented in Table S1 and Figures S1−S3 (Supporting Information). Already well documented, any intravenously injected nanoparticulate system including CNTs predominantly localize in Kupffer cells and not in heptocytes.24,25 Moreover, as already described in our previous study, in vivo administration of p/f-MWCNTs with lower degree of oxidation is associated with significant hepatotoxicity. We, therefore, sought to proceed with some cell line that can mimic liver Kupffer cells. Subsequently, the RAW 264.7 cell line was chosen as our cellular model because this cell line not only presents many common properties and functions of the liver Kupffer cells but also has been widely used as a model of liver macrophages. On the basis of our previously reported biodistribution data, we tried to make a crude estimate of the dose received by each Kupffer cells and considered administering an equivalent dose in the culture medium for in vitro cytotoxicity assessment. According to our calculations, each cell in the culture should have received at least 2.5−3.5 pg of p/f-MWCNTs to simulate the dose received by each Kupffer cell under in vivo conditions. We normalized this dose to a concentration of CNTs per unit volume of the culture medium and found that a concentration of 0.125−0.175 μg/mL of p-/fMWCNTs, at the minimum, is required to simulate the in vivo experimental conditions (see Supporting Information for detailed calculations). However, a number of reports have mentioned the tremendous differences between the administered doses in culture medium and doses associated with cells because cells respond to materials they come in contact with (delivered dose) or subsequently internalize, not to materials that they remain suspended in the media over the course of an experiment.26 Moreover, we had no presumption/intitution 2130

dx.doi.org/10.1021/tx300228d | Chem. Res. Toxicol. 2012, 25, 2127−2137

Chemical Research in Toxicology

Article

CNTs were comparable. Similar results were obtained when cytotoxicity was determined by the CB assay (See Supporting Information, Figure S5). The results, however, do not correlate with our in vivo findings. As reported in our earlier study, an increase in oxidation time from 0 to 4 h was accompanied with improved hydrophilicity and reduced hepatotoxicity; further oxidation had no significant impact on the toxicity profile of fMWCNTs. The opposing trend of in vitro and in vivo studies suggested that for any biomedical applications involving CNTs, the degree of functionalization should be carefully chosen so as to ensure maximum biocompatibility and minimal toxicity at both the organ and cellular levels. CNTs with higher functionalization densities may not influence the extracellular toxicity; however, an alteration in surface carboxyl density may severely transform the cellular responses to CNTs. These results, however, set an optimal limit for the degree of CNT oxidation (3.4 mmol carboxyl groups/g CNT) that could be used for in vivo applications while inducing the least cellular toxicity. 3.3. Functionalization Density-Dependent Changes in the Levels of Apoptotic Markers. The mechanism of cell death was further determined by monitoring changes in the levels of apoptotic markers (Annexin V and TUNEL staining) and the production of free radical viz. reactive nitrogen species (RNS) and reactive oxygen species (ROS). Even at their lowest degree of functionalization, cells incubated with oxidized MWCNTs for 24 h presented a significant (p < 0.001) increase in Annexin V (AV) and TUNEL positive cells as compared to their pristine counterparts. For apoptosis studies, we selected a 24 h time point because apoptosis markers like Annexin V can even detect the number of cells in early apoptosis in contrast to tetrazolium-based assays, which count the number of viable (living) cells and consequently give a rough idea of the number of dead cells that are not detectable in the early stage of apoptosis. As evident from Figure 2B, the number of AV (+ve) and TUNEL (+ve) cells increased proportionally with the degree of functionalization indicating that oxidized MWCNT induced apoptosis critically depends on their functionalization density. The results, however, are in line with an earlier study in which oxidized MWCNTs presented higher in vitro cytotoxicity and proapoptotic activity as compared to its pristine counterpart.13 3.4. Effect of Functionalization Density on RNS/ROS Production. Reactive nitrogen species (RNS) represent a family of antimicrobial molecules derived from nitric oxide (·NO) and superoxide (O2−) produced via the enzymatic activity of inducible nitric oxide synthase 2 (NOS2) and NADPH oxidase, respectively. Reactive nitrogen species act together with reactive oxygen species (ROS) to cumulate damage in cells, causing nitrosative stress. Having scrutinized the effect of functionalization density on the cellular viability and proapoptotic activity of oxidized MWCNTs, we sought to find out how the degree of functionalization influences the RNS/ROS generation profile of acid-functionalized MWCNTs. It was quite interesting to observe that RNS production in RAW cells increased as a function of concentration and the density of the surface carboxyl group present on MWCNTs. Nevertheless, to our surprise, MWCNTs exerting the highest toxicity presented the lowest level of nitrite production at the highest CNT concentration (Figure 3). This apparent deviation from the expected behavior may be attributed to MWCNT induced cell death, which, in turn, reduces the generation of RNS inside the cells. While RNS production exhibited a direct

Figure 3. Effect of the functionalization density of oxidized MWCNTs on endogeneous RNS production. Cells were incubated with 1−100 μg/mL CNTs oxidized for 0−6 h, and the nitrite levels in the culture supernatant were determined by the Griess reagent. Data are expressed as the mean ± SEM of 3 experiments run in parallel (n = 5/ concentration/experiment). *p < 0.05 wrt pristine MWCNTs at same concentration.

proportionality with the functionalization density of oxidized MWCNTs, ROS production was detected to be the highest in the case of pristine MWCNTs and gradually decreased with an increase in the degree of functionalization. In this connection, it is worth mentioning that ROS production remained above control levels in all types of MWCNTs (Figure 4). The

Figure 4. Representative confocal images showing ROS generation in control cells and cells treated with pristine MWCNTs or acid refluxed (1, 2, 4, and 6 h) MWCNTs. Cells were incubated for 24 h with 10 μg/mL CNTs, and ROS production was determined by the DCF assay.

functionalization density-dependent reduction in oxidative stress was further confirmed by the determination of lipid peroxidation. MDA levels also showed a negative correlation with functionalization density and supported observations in the DCF assay (see Supporting Information, Figure S6). The reduction in ROS production with increased functionalization density might have been attributed to the cellular uptake of CNTs. Before investigating the cellular uptake of MWCNTs, we hypothesized that the negative charge on oxidized CNTs repel the negatively charged plasma membrane and attributed these repulsive electrostatic forces as the underlying cause for 2131

dx.doi.org/10.1021/tx300228d | Chem. Res. Toxicol. 2012, 25, 2127−2137

Chemical Research in Toxicology

Article

Figure 5. (A) Effect of inhibitors of clathrin, caveolin, and phagocytosis on the cellular uptake of CNTs. (B) Cellular uptake of pristine and 1−6 h oxidized MWCNTs, PLL-MWCNTs, PEG-MWCNTs, and MWCNTs in the presence of dextran sulfate (CNT + DS). Cells were incubated with 100 μg/mL CNTs for 12 h, and cellular uptake was determined by spectrophotometry. Data are expressed as the mean ± SEM of 3 experiments run in parallel (n = 5/concentration/experiment). Cellular uptake is expressed as the % of CNTs added in culture medium that are internalized by cells.

the reduced cellular uptake of CNTs. Contrary to our expectations, reduction in ROS production could not be correlated with the cell uptake profile of MWCNTs. The results are detailed in the subsequent section. 3.5. In Vitro Cellular Uptake of Oxidized MWCNTs and Their Mathematical Modeling. For cell uptake studies, pristine and carboxylated CNTs with varying degrees of functionalization were labeled with 5-DTAF. In all cases, the labeling efficiency of MWCNTs was higher than 95%, suggesting the formation of a strong supramolecular complex constituting of CNTs and DTAF. The stability of DTAFMWCNTs was assessed by exposing the dye labeled CNTs in culture media over a period of 1−7 days. As evident fluorimetrically, less than 5% of the loaded dye was detached from MWCNTs even after 7 days of incubation (data not shown). These results suggested that dye labeled CNTs prepared in the course of our study were stable under biological conditions. Confocal fluorescence images of RAW cells incubated with pristine and 1−6 h oxidized CNTs have been presented in Figure S7 (A) (see Supporting Information). As evident from these images, both pristine and oxidized MWCNTs were internalized by their cellular target. However, the intensity of fluorescence signals increased as a function of the carboxyl density associated with MWCNTs indicating that the cellular uptake of CNTs was directly proportional to their functionalization density. Although confocal microscopic studies qualitatively confirmed the functionalization density dependent uptake of MWCNTs, the images lacked high resolution/sharpness, which may be a consequence of light scattering by CNTs. Having confirmed the cellular internalization of MWCNTs, we quantified their uptake using the protocol described in section 2.4.6. It was interesting to observe that carboxylated MWCNTs were internalized more efficiently by their target cells as compared to their pristine counterparts (Figure 5). Carboxyl enrichment of MWCNTs not only enhanced their cellular uptake but the internalization efficiency steadily increased with an increase in the degree of functionalization. At the highest degree of functionalization, the cellular uptake was about double that of pristine MWCNTs (Figure 5). These results clearly indicate that the production of ROS was independent of the concentration of CNTs

internalized by a specific cell population. According to a recent report by Chompoosor et al., ROS production in gold nanoparticle (AuNP) incubated HeLa cells was hardly influenced by their surface functionalization.27 It was further established by this group of authors that hydrophobicity plays an important role in ROS production. As evident from contact angle measurement (see Supporting Information, Table S1), pristine MWCNTs exhibited the highest hydrophobicity, and the latter decreased with an increased degree of surface oxidation. Hence, the most plausible explanation that might account for this reduced cellular uptake of pristine MWCNTs relative to their acid-oxidized counterparts is their inherent hydrophobicity. As the degree of functionalization increases, CNTs become more and more hydrophilic, and the level of CNT induced oxidative stresses gradually mitigates. As discussed in the preceding section, cellular uptake of oxidized MWCNTs was observed to increase proportionally with the density of surface carboxyl groups. This result seems to be quite paradoxical because the negatively charged plasma membrane is likely to repel the negatively charged CNTs and subsequently restrain their cellular internalization. In an attempt to account for such anomalous behavior of oxidized MWCNTs, we could chalk out two plausible factors that might be responsible for the observed deviation from our expected results. First, shortened, oxidized CNTs are more efficiently internalized by cells, and the size-dependent effect dominates over the electrostatic effects. Second, the CNTs may be internalized by a mechanism which is more effective for negatively charged particles. To dissect the factors influencing cellular uptake, first the charge of CNTs was neutralized by coating with a polycationic polymer (poly-L-lysine; PLL) or by covalent modification with polyethylene glycol (PEG) as described in the Materials and Methods section. Following charge neutralization with PLL, the cellular uptake of CNTs was reduced but not completely abolished, suggesting that the length of CNTs critically influences CNT uptake, albeit the contribution is relatively lower (∼25%) than charge contribution (∼75%). Similar results were observed in the case of PEGMWCNTs too (Figure 5). Further, both PLL-MWCNTs and PEG-MWCNTs were nontoxic up to 100 μg/mL (Figure 6), suggesting the involvement of negative charge on the cellular 2132

dx.doi.org/10.1021/tx300228d | Chem. Res. Toxicol. 2012, 25, 2127−2137

Chemical Research in Toxicology

Article

genistin [GEN] or 10 μg/mL chlorpromazine [CPZ]) or caveolin pathway (10 mM hydroxypropyl-β-cyclodextrin [HPβCD] or 50 μg/mL nystatin [NYS]) did not blockthe cellular uptake of CNTs. However, cellular uptake was inhibited in the presence of the macropinocytosis/phagocytosis inhibitor (1 μg/mL staurosporine [SRP]) and scavenger receptor blocker (200 μg/mL dextran sulfate [DS]; Figure 5) suggesting involvement of a scavenger receptor-mediated phagocytosis mechanism. Further, CNTs exhibit a high structural resemblance with dextran sulfate, a scavenger receptor ligand,28 indicative of their ability to bind to scavenger receptors. Among the various types of scavenger receptors, type A scavenger receptors are highly expressed on RAW 264.7 cells,28 and these receptors are selective for oxidized and acetylated ligands.29,30 Considering the high expression of these receptors and the oxidized nature of CNTs, we believe that CNTs may be internalized via type A scavenger receptors. This is also supported by the inhibition of CNT uptake by dextran sulfate, a type A scavenger receptor antagonist.31,32 The results are in agreement with earlier studies suggesting a scavenger receptormediated uptake mechanism for MWCNTs33 and negatively charged NPs.34,35 Cumulatively, these results suggest that the presence of high-density negative charge plays a crucial role in augmenting the cellular uptake of acid-oxidized MWCNTs. An increased accumulation of CNTs inside their target cells is accompanied by a concomitant decrease in cellular viability and subsequent increase in the number of apoptotic cells. Further, scavenger receptor ligands have been demonstrated to reduce ROS production through the modulation of various signaling mechanisms.36,37 An increased functionalization density is believed to strengthen the interaction and binding of MWCNTs with scavenger receptors, which in turn results in the reduced ROS production observed in the present study. The size of CNTs played a relatively minor role in toxicity and contributed to approximately 25% of the observed toxicity. This size dependent effect may arise due to the fact that reduction in CNT length may lead to an increase in number density (number of CNTs per mL of culture medium), which in turn might have augmented the interaction with cells. This increased interaction may result in physical damage to the cell surface through a nanoneedle mechanism.18 In this regard, it may be further noted that the cellular uptake of MWCNTs was studied only at 10 μg/mL whereby the uptake of negatively charged CNTs were predominantly mediated through the scavenger− receptor. However, the possibility that the saturation of the scavenger−receptor may lead to cellular uptake by other mechanisms cannot be ruled out and requires further investigations. Similarly, the involvement of nonscavenger− receptor mechanisms at low CNT concentrations and their saturation at the experimental concentration used in this study cannot be disregarded and may be put forward as a subject for future investigation. 3.6. Comparison of in Vitro and in Vivo Results. Although in vitro and in vivo systems represent two distinct biological models for toxicity evaluation, several groups have reported on the concurrence between in vitro and in vivo results,38−40 while others have argued that results of in vitro assays do not always coincide with those of in vivo tests.41,42 Even though the main focus of this work was to investigate the functionalization density dependent toxicity behavior of oxidized MWCNTs in vitro, we tried to compare the results of in vitro toxicity studies with our previously reported in vivo results. Thus, for our in vitro studies, we proceeded with exactly

Figure 6. Effect of PLL-CNTs (A) and PEG-CNTs (B) on cell viability. Cells were incubated with 1−100 μg/mL CNTs for 96 h, and cell viability was determined by the MTT assay. Data are expressed as the mean ± SEM of 3 experiments run in parallel (n = 5/ concentration/experiment).

uptake and cytotoxicity of CNTs. The combined effect of CNT length and surface charge on cellular uptake is shown in Figure 7A. As evident from the figure, the more negative the zeta potential and shorter the length of MWCNTs, the higher is their cellular uptake. Multiple linear regression analysis was performed to develop a quantitative model for the cellular uptake of CNTs. The mathematical model was solved by the following equation: cellular uptake = 8.158 − (0.006· zeta potential) − (0.180· length)

The predicted cellular uptake from the above equation showed a good correlation (concordance correlation coefficient 0.953; Pearson’s coefficient = 0.955; bias correction factor = 0.999) to experimental cellular uptake. Further, all values were within 95% confidence intervals suggesting the validity of the mathematical model (Figure 7B). Similarly, the Bland-Altman difference plot and the Passing-Bablok residual plot also showed a good correlation between the experimental and predicted cellular uptakes (Figure 8). Having established a mathematical model for length and negative charge mediated cellular uptake of CNTs, we attempted to look into the mechanism responsible for enhanced uptake of negatively charged CNTs. Quantitative cellular uptake studies showed that 30 min of preincubation of cells with inhibitors of the clathrin pathway (100 μg/mL 2133

dx.doi.org/10.1021/tx300228d | Chem. Res. Toxicol. 2012, 25, 2127−2137

Chemical Research in Toxicology

Article

Figure 7. (A) Contour plot showing the combined effect of zeta potential and CNT length on the cellular uptake of CNTs; (B) a comparison between experimental cellular uptake and cellular uptake predicted from multiple linear regression analysis; (C) structural similarity between MWCNTs and dextran sulfate.

the same material that had been used earlier in our in vivo studies. Remarkably, the observed trends of toxicity in vitro and in vivo were completely different. The results, though apparently paradoxical, can be explained if we carefully compare the experimental conditions as well as the doses selected for in vitro and in vivo studies. Under in vivo conditions, CNTs induce an initial toxic/inflammatory response due to their extracellular presence. Increased cellular uptake of smaller CNTs with higher degree of functionalization leads to lower concentrations of CNTs present extracellularly and so lower inflammation. Following a similar paradigm, carboxylated CNTs are more

readily internalized and induce lower inflammation in vivo. Further, although the RAW 264.7 cells show responses similar to that of Kupffer cells, there is no exact match of functions. This, also, may be one of the underlying causes to the observed in vitro−in vivo differencs of effects. While calculating the dose of MWCNTs received by each Kupffer cell, we approximated that 100% of the accumulated dose in the liver is distributed exclusively into Kupffer cells. However, the actual physiological scenario may be considerably different because even after reaching their cellular target, CNTs have to encounter certain extracellular barriers. Therefore, the actual dose received by 2134

dx.doi.org/10.1021/tx300228d | Chem. Res. Toxicol. 2012, 25, 2127−2137

Chemical Research in Toxicology

Article

in vivo. In either case, MWCNT induced oxidative damage/ ROS production was closely associated with other factors such as metal impurities associated with CNTs and/or their surface hydrophobicity. Overall, the opposing trends of in vitro and in vivo toxicity do not completely abandon the feasibility of using in vitro models for the prediction of in vivo toxicity but discourages the extrapolation of in vitro results to in vivo.

4. CONCLUSIONS In conclusion, we have studied the functionalization density dependence of MWCNTs’ toxicity in RAW 264.7 macrophages and tried to correlate the results with our previously reported in vivo findings. Functionalization density dependence of CNTs’ toxicity in vitro showed an exactly opposite trend when compared and contrasted with in vivo results. Our studies indicated that the cellular toxicity of CNTs increases as a function of their surface carboxyl density. Mechanistic studies further suggested that the cellular uptake of oxidized CNTs was primarily mediated through scavenger receptors with a small contribution by reduced CNT length. Functionalization density dependent enhanced cellular uptake was accompanied by increased intracellular stress response as well as cytotoxicity, which in turn, resulted in cell death. Of note, oxidized MWCNT induced cytotoxicity could be effectively alleviated by PEGylation of the free carboxyl groups, which indicated that negative charges associated with CNTs’ surface had a distinct role in their mechanistic pathway of cellular internalization. Overall, these findings provide some fundamental information on the effect of oxidation as well as the degree of surface carboxylation on the in vitro cytotoxicity, free radical generation, and cellular uptake of acid-oxidized CNTs, which might be beneficial for their future biomedical applications as an attempt to compare the in vitro and in vivo toxicity behavior of oxidized CNTs revealed completely opposing trends. The results were explained on the basis of the different experimental conditions chosen and the differences between administered and actual doses associated with cells under in vitro and in vivo conditions. Small contributions by physical damage to the cell membrane were also taken to consideration. The opposing trends of in vitro and in vivo studies do not entail that in vitro models should be abandoned but discourage the extrapolation of in vivo behavior of CNTs as well as any other functional nanocarrier merely on the basis of in vitro studies. Finally, it requires to be noted that the toxicity of CNTs not only depends on their functionalization density but also on many other factors, which include their physical form, length, diameter, and the nature of functional molecules attached onto their surface. These considerations should be persistently kept in mind and revisited frequently during the development of any CNT-based nanomedical devices.

Figure 8. Statistical validation of the mathematical model with experimental cellular uptake. (A) Bland-Altman difference plot with line of equality and 95% CI; (B) Passing-Bablok residual plot.

each Kupffer cell in vivo may be even less than the theoretically calculated dose to be administered in culture medium for the simulation of in vivo effects. Thus, even if we normalize our in vivo doses to the administered dose in cell culture in vitro, it appears that cultured cells will be exposed to a much higher concentration of CNTs than that received by an equivalent population Kupffer cells in vivo. For in vitro experiments, we used a concentration ranging from 0.1 to 100 μg/mL. As evident from our cytotoxicity data, both p- and f-MWCNTs induced severe cytotoxicity at 10 μg/mL. According to our calculation, this dose corresponded to approximately 0.2 ng of MWCNTs/cell, respectively, which is approximately 60−80 times higher than the estimated theoretical dose. However, as already discussed in section 3.1, 100% of the administered doses are never taken up by cells. As evident from cell uptake studies, less than 10% of administered CNTs were taken up by RAW cells. In that case, the approximate dose associated with each cell is ∼0.02 ng, which is still 6−8 times higher than the theoretically calculated dose associated with each Kupffer cells under in vivo conditions. Thus, the higher the surface carboxyl density of MWCNTs, more will be their intracellular concentration and subsequent higher toxicity of the test material. It was, however, interesting to observe that the opposing trends of in vitro and in vivo toxicity are mostly evident at higher concentrations (>1 μg/mL). At lower concentrations (∼0.1 μg/mL), irrespective of their degree of functionalization, all MWCNT preparations presented very negligible cytotoxicity. It appears that at this low concentration, intracellular concentrations of MWCNTs are too low to induce any cytotoxic response. Another point which deserves mention in this regard is that despite the opposing trends of in vitro and in vivo toxicity, p/f-MWCNT induced oxidative damage was independent of their functionalization density both in vitro and

■

ASSOCIATED CONTENT

S Supporting Information *

Calculation of dose received by each Kupffer cells in vivo and normalization to in vitro dose in cell culture, scanning electron micrographs, FTIR spectra, a TG thermogram, effect of the functionalization density of oxidized MWCNTs on the viability of RAW 264.7 cells and on lipid peroxidation, representative fluorescence images, and particle characteristics of acid-oxidized CNTs. This material is available free of charge via the Internet at http://pubs.acs.org. 2135

dx.doi.org/10.1021/tx300228d | Chem. Res. Toxicol. 2012, 25, 2127−2137

Chemical Research in Toxicology

■

Article

(11) Ji, Z., Zhang, D., Li, L., Shen, X., Deng, X., Dong, L., Wu, M., and Liu, Y. (2009) The hepatotoxicity of multi-walled carbon nanotubes in mice. Nanotechnology 20, 445101. (12) Patlolla, A. K., Hussain, S. M., Schlager, J. J., Patlolla, S., and Tchounwou, P. B. (2010) Comparative study of the clastogenicity of functionalized and nonfunctionalized multiwalled carbon nanotubes in bone marrow cells of Swiss Webster mice. Environ. Toxicol. 25, 608− 621. (13) Ryman-Rasmussen, J. P., Riviere, J. E., and Monteiro-Riviere, N. A. (2007) Surface coatings determine cytotoxicity and irritation potential of quantum dot nanoparticles in epidermal keratinocytes. J. Invest. Dermatol. 127, 143−153. (14) Lacerda, L., Ali-Boucetta, H., Herrero, M. A., Pastorin, G., Bianco, A., Prato, M., and Kostarelos, K. (2008) Tissue histology and physiology following intravenous administration of different types of functionalized multiwalled carbon nanotubes. Nanomedicine 3, 149− 161. (15) Sayes, C. M., Liang, F., Hudson, J. L., Mendez, J., Guo, W., Beach, J. M., Moore, V. C., Doyle, C. D., West, J. L., and Billups, W. E. (2006) Functionalization density dependence of single-walled carbon nanotubes cytotoxicity in vitro. Toxicol. Lett. 161, 135−142. (16) Jain, S., Thakare, V. S., Das, M., Godugu, C., Jain, A. K., Mathur, R., Chuttani, K., and Mishra, A. K. (2011) Toxicity of multiwalled carbon nanotubes with end defects critically depends on their functionalization density. Chem. Res. Toxicol. 24, 2028−2039. (17) Tejral, G., Panyala, N. R., and Havel, J. (2009) Carbon nanotubes: toxicological impact on human health and environment. J. Appl. Biomed. 7, 1−13. (18) Contera, S. A., Trigueros, S., and Ryan, J. F. (2009) Nanotubes as drug delivery systems for prokaryotic and eukaryotic cells. Biophys. J. 96, 51a. (19) Swarnakar, N. K., Jain, A. K., Singh, R. P., Godugu, C., Das, M., and Jain, S. (2011) Oral bioavailability, therapeutic efficacy and reactive oxygen species scavenging properties of coenzyme Q10loaded polymeric nanoparticles. Biomaterials 32, 6860−6874. (20) Jang, J., Lim, D.-H., and Choi, I.-H. (2010) The impact of nanomaterials in immune system. Immune Network 10, 85−91. (21) Tsikas, D. (2005) Review Methods of quantitative analysis of the nitric oxide metabolites nitrite and nitrate in human biological fluids. Free Radical Res. 39, 797−815. (22) Ryman-Rasmussen, J. P., Riviere, J. E., and Monteiro-Riviere, N. A. (2007) Variables influencing interactions of untargeted quantum dot nanoparticles with skin cells and identification of biochemical modulators. Nano Lett. 7, 1344−1348. (23) Jain, S., Thakare, V. S., Das, M., Godugu, C., Jain, A. K., Mathur, R., Chuttani, K., and Mishra, A. K. (2011) Toxicity of multiwalled carbon nanotubes with end defects critically depends on their functionalization density. Chem. Res. Toxicol. 24, 2028−2039. (24) Miyawaki, J., Matsumura, S., Yuge, R., Murakami, T., Sato, S., Tomida, A., Tsuruo, T., Ichihashi, T., Fujinami, T., and Irie, H. (2009) Biodistribution and ultrastructural localization of single-walled carbon nanohorns determined in vivo with embedded Gd2O3 labels. ACS Nano 3, 1399−1406. (25) Goncalves, C., Torrado, E., Martins, T., Pereira, P., Pedrosa, J. Gama, M. Dextrin nanoparticles: Studies on the interaction with murine macrophages and blood clearance. Colloids Surf., B 75, 483-489. (26) Teeguarden, J. G., Hinderliter, P. M., Orr, G., Thrall, B. D., and Pounds, J. G. (2007) Particokinetics in vitro: dosimetry considerations for in vitro nanoparticle toxicity assessments. Toxicol. Sci. 95, 300−312. (27) Chompoosor, A., Saha, K., Ghosh, P. S., Macarthy, D. J., Miranda, O. R., Zhu, Z. J., Arcaro, K. F., and Rotello, V. M. (2010) The role of surface functionality on acute cytotoxicity, ROS generation and DNA damage by cationic gold nanoparticles. Small 6, 2246−2249. (28) Lee, H. A., Imran, M., Monteiro-Riviere, N. A., Colvin, V. L., Yu, W. W., and Riviere, J. E. (2007) Biodistribution of quantum dot nanoparticles in perfused skin: evidence of coating dependency and periodicity in arterial extraction. Nano Lett. 7, 2865−2870.

AUTHOR INFORMATION

Corresponding Author

*Tel: +91-172-2292055. Fax: +91-172-2214692. E-mail: [email protected]. Funding

This study was funded by grants from Indian Council of Medical Research (ICMR), Government of India. M.D. and R.P.S. received Post Doctoral and Senior Research Fellowships, respectively, from Department of Science and Technology, Government of India. Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS We acknowledge Mr. Dinesh Chauhan, Mr. B. Mallikarjun, and Mr. Manish for technical assistance. ABBREVIATIONS CNTs, carbon nanotubes; DMEM, Dulbecco’s modified Eagle’s medium; FBS, fetal bovine serum; MWCNTs, multiwalled carbon nanotubes; f-MWCNTs, functionalized multiwalled carbon nanotubes; NPs, nanoparticle; RNS, reactive nitrogen species; ROS, reactive oxygen species

■

REFERENCES

(1) Zhu, X.-D., Zhuang, Y., Ben, J.-J., Qian, L.-L., Huang, H.-P., Bai, H., Sha, J.-H., He, Z.-G., and Chen, Q. (2011) Caveolae-dependent endocytosis is required for class a macrophage scavenger receptormediated apoptosis in macrophages. J. Biol. Chem. 286, 8231−8239. (2) Jain, A. K., Swarnakar, N. K., Das, M., Godugu, C., Singh, R. P., Rao, P. R., and Jain, S. (2011) Augmented anticancer efficacy of doxorubicin loaded polymeric nanoparticles after oral administration in breast cancer induced animal model. Mol. Pharmaceutics 8, 1140− 1151. (3) Jain, S., Mathur, R., Das, M., Swarnakar, N. K., and K., M. A. (2011) Synthesis, pharmacoscintigraphic evaluation and antitumor efficacy of methotrexate loaded, folate conjugated, stealth albumin nanoparticles. Nanomedicine 6, 1733−1754. (4) Gil, P. R., Oberdorster, G., Elder, A., Puntes, V., and Parak, W. J. (2010) Correlating physico-chemical with toxicological properties of nanoparticles: the present and the future. ACS Nano 4, 5527−5531. (5) Kolosnjaj-Tabi, J., Hartman, K. B., Boudjemaa, S., Ananta, J. S., Morgant, G., Szwarc, H., Wilson, L. J., and Moussa, F. (2010) In vivo behavior of large doses of ultrashort and full-length single-walled carbon nanotubes after oral and intraperitoneal administration to swiss mice. ACS Nano 4, 1481−1492. (6) Das, M., Bandyopadhyay, D., Mishra, D., Datir, S., Dhak, P., Jain, S., Maiti, T. K., Basak, A., and Pramanik, P. (2011) Clickable, trifunctional magnetite nanoparticles and their chemoselective biofunctionalization. Bioconjugate Chem. 22, 1181−1193. (7) Das, M., Mishra, D., Maiti, T. K., Basak, A., and Pramanik, P. (2008) Bio-functionalization of magnetite nanoparticles using an aminophosphonic acid coupling agent: new, ultradispersed, iron-oxide folate nanoconjugates for cancer-specific targeting. Nanotechnology 19, 415101. (8) Bianco, A., Kostarelos, K., and Prato, M. (2008) Opportunities and challenges of carbon-based nanomaterials for cancer therapy. Expert Opin. Drug Delivery 5, 331−342. (9) Petushkov, A., Intra, J., Graham, J. B., Larsen, S. C., and Salem, A. K. (2009) Effect of crystal size and surface functionalization on the cytotoxicity of silicalite-1 nanoparticles. Chem. Res. Toxicol. 22, 1359− 1368. (10) Yang, S. T., Wang, H., Meziani, M. J., Liu, Y., Wang, X., and Sun, Y. P. (2009) Biodefunctionalization of functionalized single-walled carbon nanotubes in mice. Biomacromolecules 10, 2009−2012. 2136

dx.doi.org/10.1021/tx300228d | Chem. Res. Toxicol. 2012, 25, 2127−2137

Chemical Research in Toxicology

Article

(29) Itabe, H., Obama, T., and Kato, R. (2011) The dynamics of oxidized LDL during atherogenesis. J. Lipids, DOI: 10.1155/2011/ 418313. (30) Tamura, Y., Osuga, J.-i., Adachi, H., Tozawa, R.-i., Takanezawa, Y., Ohashi, K., Yahagi, N., Sekiya, M., Okazaki, H., Tomita, S., Iizuka, Y., Koizumi, H., Inaba, T., Yagyu, H., Kamada, N., Suzuki, H., Shimano, H., Kadowaki, T., Tsujimoto, M., Arai, H., Yamada, N., and Ishibashi, S. (2004) Scavenger receptor expressed by endothelial cells I (SREC-I) mediates the uptake of acetylated low density lipoproteins by macrophages stimulated with lipopolysaccharide. J. Biol. Chem. 279, 30938−30944. (31) Thelen, T., Hao, Y., Medeiros, A. I., Curtis, J. L., Serezani, C. H., Kobzik, L., Harris, L. H., and Aronoff, D. M. (2010) The class A scavenger receptor, macrophage receptor with collagenous structure, is the major phagocytic receptor for Clostridium sordellii expressed by human decidual macrophages. J. Immunol. 185, 4328−4335. (32) Limmon, G. V., Arredouani, M., McCann, K. L., Corn Minor, R. A., Kobzik, L., and Imani, F. (2008) Scavenger receptor class-A is a novel cell surface receptor for double-stranded RNA. FASEB J. 22, 159−167. (33) Campa, V. M., Iglesias, J. M., Carcedo, M. T., Rodríguez, R., Riera, J., Ramos, S., and Lazo, P. S. (2005) Polyinosinic acid induces TNF and NO production as well as NF-kB and AP-1 transcriptional activation in the monocytemacrophage cell line RAW 264.7. Inflammation Res. 54, 328−337. (34) Kanno, S., Furuyama, A., and Hirano, S. (2007) A murine scavenger receptor MARCO recognizes polystyrene nanoparticles. Toxicol. Sci. 97, 398−406. (35) Plourde, N., Kortagere, S., Welsh, W., and Moghe, P. (2009) Structure activity relations of nanolipoblockers with the atherogenic domain of human macrophage scavenger receptor A. Biomacromolecules 10, 1381−1391. (36) Ying, E., and Hwang, H. (2010) In vitro evaluation of the cytotoxicity of iron oxide nanoparticles with different coatings and different sizes in A3 human T lymphocytes. Sci. Total Environ. 408, 4475−4481. (37) Shnyra, A., and Lindberg, A. A. (1995) Scavenger receptor pathway for lipopolysaccharide binding to Kupffer and endothelial liver cells in vitro. Infect. Immun. 63, 865−873. (38) Gunnison, A. F. (1981) Sulphite toxicity: A critical review of in vitro and in vivo data. Food Cosmet. Toxicol. 19, 667−682. (39) Piersma, A. H., Janer, G., Wolterink, G., Bessems, J. G. M., Hakkert, B. C., and Slob, W. (2008) Quantitative extrapolation of in vitro whole embryo culture embryotoxicity data to developmental toxicity in vivo using the benchmark dose approach. Toxicol. Sci. 101, 91−100. (40) Williams, C. S., Watson, A. J. M., Sheng, H., Helou, R., Shao, J., and DuBois, R. N. (2000) Celecoxib prevents tumor growth in vivo without toxicity to normal gut: lack of correlation between in vitro and in vivo models. Cancer Res. 60, 6045. (41) Schapira, A., Bygbjerg, I. C., Jepsen, S., Flachs, H., and Bentzon, M. W. (1986) The susceptibility of Plasmodium falciparum to sulfadoxine and pyrimethamine: correlation of in vivo and in vitro results. Am. J. Trop. Med. Hyg. 35, 239. (42) Johnson, J. I., Decker, S., Zaharevitz, D., Rubinstein, L. V., Venditti, J. M., Schepartz, S., Kalyandrug, S., Christian, M., Arbuck, S., and Hollingshead, M. (2001) Relationships between drug activity in NCI preclinical in vitro and in vivo models and early clinical trials. Br. J. Cancer 84, 1424.

2137

dx.doi.org/10.1021/tx300228d | Chem. Res. Toxicol. 2012, 25, 2127−2137