Toxicology Letters 237 (2015) 61–71

Contents lists available at ScienceDirect

Toxicology Letters journal homepage: www.elsevier.com/locate/toxlet

Deciphering the underlying mechanisms of oxidation-state dependent cytotoxicity of graphene oxide on mammalian cells Wendi Zhanga,1, Liang Yanb,1, Meng Lic, Ruisheng Zhaob , Xiao Yanga , Tianjiao Jia , Zhanjun Gub , Jun-Jie Yinc , Xingfa Gaob , Guangjun Niea,* a

Key Laboratory for Biomedical Effects of Nanomaterials & Nanosafety, National Center for Nanoscience and Technology, Beijing 100190, China Key Laboratory for Biomedical Effects of Nanomaterials and Nanosafety, Institute of High Energy Physics, Chinese Academy of Sciences, Beijing 100049, China c Division of Analytical Chemistry, Office of Regulatory Science, Center for Food Safety and Applied Nutrition, U.S. Food and Drug Administration, College Park, MD 20740, USA b

H I G H L I G H T S

G R A P H I C A L A B S T R A C T


A strong association between the oxidation level and toxicity of GO was found.
A systemic chemical mechanism of GO cytotoxicity is proposed.
ROS conversion facilitation and the oxidative ability of GO are attributable.
The key role of carboxyl groups in varying the energy barrier of ROS reaction.
Both experiments and theoretical simulation were conducted.

The cytotoxicity of three GOs with different oxidation degrees, GO-h, m and l (GO-high, medium and low) were tested on mouse embryo fibroblasts (MEFs). Significantly higher level of oxidative stress in cells was induced by the GO with lower oxidation degree in relation to its stronger toxicity to cells. This can be attributed to its stronger indirect oxidative damages through facilitating ROS conversion and higher direct oxidative abilities on intracellular biomolecules. Theoretical calculation indicated the key contributions of oxygenation-level-based nanostructure to varying the energy barrier of H2O2 decomposition reaction.

A R T I C L E I N F O

A B S T R A C T

Article history: Received 22 March 2015 Received in revised form 29 May 2015 Accepted 31 May 2015 Available online 3 June 2015

The promising broad applications of graphene oxide (GO) derivatives in biomedicine have raised concerns about their safety on biological organisms. However, correlations between the physicochemical properties, especially oxidation degree of GOs and their toxicity, and the underlying mechanisms are not well understood. Herein, we evaluated the cytotoxicity of three GO samples with various oxidation degrees on mouse embryo fibroblasts (MEFs). Three samples can be internalized by MEFs observed via transmission electron microscopy (TEM), and were well tolerant by MEFs at lower doses (below 25 mg/ ml) but significantly toxic at 50 and 100 mg/ml via Cytell Imaging System. More importantly, as the oxidation degree decreased, GO derivatives led to a higher degree of cytotoxicity and apoptosis. Meanwhile, three GOs stimulated dramatic enhancement in reactive oxygen species (ROS) production in MEFs, where the less oxidized GO produced a higher level of ROS, suggesting the major role of oxidative stress in the oxidation-degree dependent toxicity of GOs. Results from electron spin resonance (ESR) spectrometry showed a strong association of the lower oxidation degree of GOs with their stronger indirect oxidative damage through facilitating H2O2 decomposition into
OH and higher direct oxidative

Key words: Cytotoxicity Graphene oxide (GO) Reactive oxygen species (ROS) Apoptosis Electron spin resonance (ESR) spectrometry

* Corresponding author at: National Center for Nanoscience and Technology, Chinese Academy of Science, No.11, Beiyitiao Zhongguancun, Beijing 100190, China. Tel.: +86 10 8254 5529 E-mail address: [email protected] (G. Nie). 1 These authors contributed equally to this work. http://dx.doi.org/10.1016/j.toxlet.2015.05.021 0378-4274/ ã 2015 Elsevier Ireland Ltd. All rights reserved.

62

W. Zhang et al. / Toxicology Letters 237 (2015) 61–71

abilities on cells. The theoretical simulation revealed the key contributions of carboxyl groups and aromatic domain size of nanosheets to varying the energy barrier of H2O2 decomposition reaction. These systematic explorations in the chemical mechanisms unravel the key physicochemical properties that would lead to the diverse toxic profiles of the GO nanosheets with different oxygenation levels, and offer us new clues in the molecular design of carbon nanomaterials for their safe applications in biomedicine. ã2015 Elsevier Ireland Ltd. All rights reserved.

1. Introduction The potential nanotoxicity of graphene and its derivatives (such as graphene oxide, GO) has received increasing attention due to their great range of potential applications, especially in biomedical fields, such as biosensors (Akhavan et al., 2012a), photothermal therapy (Sun et al., 2008) and drug delivery (Zhang et al., 2011). Although various covalent and non-covalent methods for GO functionalization have improved their stability and biocompatibility, and even reduced their toxicity in physiological environments (An et al., 2013; Liu et al., 2010; Pan et al., 2012), GO itself still occupies the most important position in the initial molecular design process. We have to face the inevitable issue to what oxidation degree we should “oxidize” graphite, the initial source to synthesize GO. Therefore, to evaluate the safety of GO on biological organisms, unraveling the association between the toxicity and the oxidation degree of nanosheets becomes an urgent task. So far there have been a number of studies regarding the biological effects of GO and reduced GO (rGO) (or pristine graphene) on models like mammalian cells (Das et al., 2013; Sasidharan et al., 2011) and bacteria (such as E. coli) (Liu et al., 2011; Akhavan and Ghaderi, 2010) as well as mice (Wang et al., 2011) and C. elegans (Zhang et al., 2012). However, the researchers have not reached to consensus on the correlation between oxidation degree of nanosheets and their toxicity. More importantly, the previous reports regarding chemical mechanisms underlying the toxicity of GO samples with different oxygenated functionalization are still limited (Das et al., 2013; Liu et al., 2011, 2012). One point held that GO is more toxic than rGO. As for mammalian cell models, GO was found to be more toxic than rGO on human umbilical vein endothelial cells (HUVEC) (Das et al., 2013). It was hypothesized that more reactive functional groups (e.g. OH, COOH) of GO would have a greater potential to interact with biological macromolecules compared with reduced GO, emphasizing the favor of oxygenated groups on enhancing the bio-nano interaction efficiency. As for bacterial models, both membrane damage and oxidative stress were proposed to be critical to GO toxicity (Liu et al., 2011), and the smaller size and better dispersion of more oxidized GO in physiological conditions might be responsible for their higher antibacterial activity on E. coli. Contrarily, it has also been reported that rGO processed higher toxicity than GO on mammalian cells (Sasidharan et al., 2011) and bacteria (Akhavan and Ghaderi, 2010). Sasidharan et al. (2011) reported that large accumulation of pristine graphene on the monkey kidney cell membrane induces intracellular reactive oxygen species (ROS) production thus leading to apoptosis, while carboxyl functionalized graphene allowed cells to function relatively normally, though internalized by the cells. Akhavan and Ghaderi (2010) proposed that the higher antibacterial effects of the reduced GO might be attributed to the better charge transfer between the bacteria and the more sharpened edges of the reduced GO during the contact interaction. Besides, some researchers also found both GO and rGO solid substrates were biocompatible to NIH-3T3 cells (Ryoo et al., 2010). The discrepancy may be attributed to different sample properties and biological models. In this work, we synthesized three GO samples, GO-h, GO-m and GO-l (GO-high, GO-medium and GO-low), all with good

solubility in water and similar sizes but distinct oxidation levels. By this we aimed to minimize the interference of size and solubility when evaluating the correlation between the oxidation state and the toxicity of GO. We investigated their effects on the internalization, morphology, viability and apoptosis of mouse embryo fibroblasts (MEFs) in relation to both the intracellular and in vitro ROS generation by GOs. Based on results from electron spin resonance (ESR) spectrometry and theoretical simulation, we present a mechanism showing a strong correlation between the oxidation-degree-based nanostructures of GO and their cytotoxicity, and for the first time interpret the chemical mechanisms involved in the different toxicological behaviors of three GO samples with distinct properties on oxidation degree. 2. Materials and methods 2.1. Preparation of GO with various degrees of oxidation Natural graphite was purchased from Alfa Aesar. KMnO4, concentrated H2SO4 and H2O2 (30%) were obtained from Beijing Chemical Reagent Co. All the chemicals were of analytical grade and used without further purification. Deionized water was used throughout the experiment. Graphite oxides with various degrees of oxidation were prepared from graphite flake using the reported modified Hummer’s method (Hummers and Offeman, 1958; Yan et al., 2013). In brief, 0.2 g of graphite was stirred in 10 ml of concentrated H2SO4 for 5 h. The required amount of KMnO4 was then gradually added to the above solution in ice-bath while keeping the temperature at less than 4  C. The obtained mixture was stirred at 37  C for 2 h. Afterwards, 20 ml of deionized water was added under vigorous stirring to dilute the resulting solution. The suspension was further treated by adding H2O2 solution (10 ml) to remove the residual KMnO4 and MgO2. The acid washed paste was filtered and washed with 10% HCl solution and deionized water until the pH value of the solution became neutral. Finally, the resulting graphite oxide was subjected to dialysis for 7 days to remove residual metal ions and acid. The cut off size of the dialysis membrane was 3500 D and the medium used was deionized water (18.2 MV/cm). The dialysis medium was refreshed every 3 h for the first day and daily for the last 6 days. The GO nanosheets were obtained by diluting the resulting graphite oxide suspension with deionized water until the concentration reached to 1 mg/ml, and then the above suspension was probe-sonicated at the power of 325 W for 4 h, followed by centrifuging at 1,760 g for 20 min to remove any unexfoliated graphite oxide. The degree of oxidation was tuned by changing the amount of KMnO4 from 0.3 g to 0.7 g with an increment of 0.2 g per oxidation level while other parameters were kept constant in the reaction. Before use, three GO liquid samples (dispersed in water) were sonicated using an ultrasonic cleaner (KQ2200E) at 100 W for 15 min. 2.2. Characterization of GO samples Atomic force microscopy (AFM) images were taken on a scanning probe microscopy (SPM, Multimode 8, Bruke) under ambient conditions. The micro-Raman spectroscopy (Renishawin Via Raman Spectroscope) experiments were performed under

W. Zhang et al. / Toxicology Letters 237 (2015) 61–71

ambient conditions with 633 nm excitation from an argon ion laser. The C1s spectra of X-ray photoelectron spectroscopy (XPS) (Thermo Scientific ESCALAB 250Xi) was fitted using a Gaussian–Lorentzian peak shape after Shirley background correction. Ultraviolet/Visible (UV–vis) spectrophotometer (TU-1901 Beijing Purkinje General Instrument Co., Ltd.) was used for UV–vis measurements. 2.3. Cell culture The MEF cells were bought from the School of Basic Medicine, Peking Union Medical College. Cells were maintained at 37  C (5% CO2) in DMEM medium (WISENT Inc.) supplemented with 10% (vol/vol) fetal bovine serum (WISENT Inc.), 2 mM L-glutamine, 20 mM HEPES, 100 U/ml penicillin and 1 mg/ml streptomycin. 2.4. Observation of MEF cells via transmission electron microscopy (TEM) MEF cells were plated in 6-well plates (5  104 cells per well) and incubated for 24 h. GO-h, GO-m and GO-l (50 mg/ml) were introduced separately to cells with a predetermined concentration in culture medium. The internalization of the GOs by MEFs was observed via TEM. Briefly, after 24 h exposure, the cells were washed with cold PBS for three times. After the centrifugation at 179g for 5 min, cells were collected, prefixed with 2.5% glutaraldehyde, post-fixed in 1% osmium tetroxide, dehydrated in a graded alcohol series, embedded in epoxy resin, and cut with an ultra microtome. Thin sections post stained with uranyl acetate and lead citrate were inspected with TEM (HT 7700).

63

measure the mean fluorescence of the intracellular reduced DCFDA molecules, which is directly related to the intracellular ROS level. 2.8. ESR experiments The sample solution contained 0.2 mg/ml different GO samples (GO-l, GO-m or GO-h) in pH 7.4 PBS buffer (100 mM). To explore the
OH production induced by GO in presence of H2O2, the reaction mixture of spin label BMPO (5-tertbutoxycarbonyl 5-methyl-1pyrroline N-oxide) (purchased from Bioanalytical Labs, Sarasota, FL) and GO was added with H2O2 (5 mM) and then transferred to a 50 mL quartz capillary tube and placed into the microwave cavity of a Bruker EMX ESR Spectrometer (Billerica, MA) for ESR spectra detection. 1-Hydroxy-3-carboxy-2,2,5,5-tetramethylpyrrolidine hydrochloride (CPH) were obtained from Enzo Life Sciences (Farmingdale, NY). The spin probe CPH was mixed with GO solutions to determine the overall oxidative ability of the sample. The ESR spectra were recorded at 20 mW microwave power, 1 G field modulation, and 100 G scan width. Distilled deionized water (18.2 MV cm) from a Milli-Q water purification system was used in all experiments. 2.9. Theoretical calculation All the theoretical calculations were performed with Gaussian 09 program (Frisch et al., 2010). The geometries were optimized at B3LYP/6-31G(d,p) level of theory in water, and the liquid environment was modeled by utilizing the Tomasi’s Polarized Continuum Model (PCM).

2.5. MEF cell viability 2.10. Statistical analysis MEFs were seeded in 96-well plates at 5  103 cells per well and cultured for 24 h before GOs (12.5, 25, 50 and 100 mg/ml) were introduced into each well at their final concentration in DMEM. After treated with the three GOs, MEFs were washed three times with PBS and added with new DMEM containing Reagent DAPI/PI of the Cytell Cell Viability Kit (code number 29-0574-96). The cells were then cultured for 45 min at 37  C in incubator before being transferred into Cytell Cell Imaging System (code number 290567-49). The cell morphological images were snapped and cell survival rate was automatically calculated by the system. 2.6. MEF cell apoptosis Two milliliter cell suspension (density 5  104/ml) was incubated in a 6-well Plate 24 h before the medium was replaced with fresh medium with or without 50 mg/ml GO. After cultured for 24 h, cells were washed three times with PBS. Then a kit (FITC Annexin V Apoposis Detection Kit I, BD PharmingenTM) containing PI and FITC conjugated-Annexin V was used according to the manufacturer’s instructions. The levels of apoptosis were assayed by flow cytometry (BD Accuri C6). 2.7. Intracellular ROS production Two milliliter cell suspension (density 5  104/ml) was dispersed in a 6-well plate for flow cytometry. The cells were preincubated for 24 h in a humidified incubator (37  C, 5% CO2). Then the medium was replaced with fresh medium (control) or GOcontaining media at 50 mg/ml and cells were treated for 24 h. Cells were washed three times with PBS and incubated with 10 mM H2DCF-DA (20 ,70 -dichlorofluorescin diacetate) (Sigma, C6883) diluted in PBS for 30 min (Kampkotter et al., 2007). Then the cells were washed three times with PBS to eliminate the excess of unreacted probe. Finally flow cytometry (BD Accuri C6) was used to

Unless mentioned otherwise, the multi-comparisons in this paper were done using One-Way ANOVA with SPSS software 16.0. 3. Results 3.1. Characterization of GO derivatives In this work we synthesized three GO samples with various degrees of oxidation, GO-h, GO-m and GO-l. The morphology of GOs was characterized using AFM. AFM analysis showed that the thickness (Fig. S1) and average size of GO nanosheets (Fig. 1A) with various oxidation degrees were 1.2 nm and 200–260 nm (n = 100), respectively, which reveals that most of GO nanosheets were single-layered and have similar size distribution. The resulting GO samples were all well dispersed in water (Fig. S2). These results are favor to evaluate the effect of oxidation degree of GO nanosheets on their toxicity. The C1s spectra of the three GOs were determined by XPS. The C/ O ratios for GO-h, GO-m and GO-l are 2.2, 2.4 and 2.9, respectively (Fig. S3). On the other hand, from GO-l, GO-m to GO-h (Fig. 1B), the ascending area proportion of the peaks at 286.9 and 288.5 eV in each spectrum, which is due to the contributions from C O and C¼O, respectively, can be also assigned to the higher amount of oxidation groups. Raman spectroscopy is a standard non-destructive tool for the characterization of structural elucidation of GO. In the Raman spectra (Fig. 1C), the D band has a higher intensity, but the intensity of 2D band decreased, which can be attributed to the formation of defects (including oxygenated functional groups) due to the oxidation reaction and the breaking of the stacking order. Also, the ratios of ID/IG, which respect the oxidation level, increased with higher degree of oxidation, suggesting that at higher oxidation level, more oxygenated functional groups at the basal plane and also at the edges were formed. In addition, from GO-h,

64

W. Zhang et al. / Toxicology Letters 237 (2015) 61–71

GO-m to GO-l, the maximum absorption peaks of the UV spectra of the GOs exhibit red shift (from 231 nm, 235 nm to 237 nm, respectively) (Fig. S4), indicating that the oxidation degree of the GOs decreased and the electronic conjugation was restored (Wu et al., 2013). These results demonstrate that from GO-l, GO-m to GO-h, the degree of oxidation gradually increases with raising the amount of KMnO4 during synthesis process. 3.2. Cell uptake and morphology In this work, MEFs were used as the toxicological model (Luo et al., 2014). Firstly, we observed the uptake behavior of GO samples by MEF cells via TEM (Fig. 2A). GO nanosheets aggregated into needle-shaped structures under TEM as indicated by Lammel et al. (2013) and distributed both on the surface and inside the cells. GO nanosheets with lateral sizes from 200 nm to 300 nm can be recognized. In contrast, there are no such structures formed in control groups. The formation of membrane invaginations can also be found on the surface the MEFs treated with GO-m (Fig. 2A, GOm, middle). The images show GO aggregates were trapped inside MEFs in the vesicles (endosomes or lysosomes). These suggest that three GO samples can all be internalized by MEFs. The ability of internalization of GO nanosheets has been also observed in mammalian cells by other researchers previously (Lammel et al., 2013). As one important indicator of cell status, the cell morphological changes after GO exposure were recorded to demonstrate the effect of GO samples on MEFs directly. In GO-h treated group (Fig. 2B), most cells were still in normal spindle- or star-shape, while in GO-l group, most cells were irregular shaped and cell size were generally smaller accompanied by a large number of round cells and cell fragments. These results reveal that GO sample with lower oxidation level has a stronger adverse effect on cell morphology.

3.3. Effects of GO on MEFs viability and apoptosis To estimate the toxicity of the GOs on MEFs, we stained MEFs with DAPI and PI to label the nucleus of all cells and dead cell, respectively, and detected cell viability by Cytell Cell Imaging System. In Fig. 3A, we can see that cell survival rate reduced down to 65% and 29% relative to control group after treatment with GOm and GO-l at 100 mg/ml for 24 h, respectively, while the rate of GO-h treated group remained at 85%. Toxicity of three GO samples at 50 mg/ml was also statistically significant compared with the control and followed the same tendency as those at 100 mg/ml. However, MEFs were well tolerant to three GO samples at lower doses of 25 and 12.5 mg/ml. Then apoptosis was tested for GO treated MEFs via flow cytometry. We found that MEF cells pretreated with the GOs (50 mg/ml) for 24 h underwent early apoptosis (Q4) (Fig. 3B and C). It can be seen that with the decreased degrees of oxidation, the proportion of cells in early apoptosis (or early plus late apoptosis) rises (p value < 0.05) while that in normal status reduces (p value < 0.05). These results indicate that the cytotoxicity of GO ascends as GO oxidation degree descends. During the early process of apoptosis, cell will experience shrinkage and the cell outlines become irregular, which are visible by light microscopy (Kerr et al., 1972; Elmore, 2007). With cell shrinkage, the cells are smaller in size. Considering the results from both morphology and apoptosis, the generally smaller sized and irregular shaped cells in Fig. 2B can be presumed to be experiencing early apoptosis process. 3.4. Intracellular ROS generation The results above from morphology, viability and apoptosis indicate that GO toxicity enhanced with the reduction of oxidation degree of the GOs. It is noteworthy that GO may possess electron transfer activity (Liu et al., 2011; Zuo et al., 2009). To explore the mechanisms of this oxidation-degree dependent toxicity of the GOs, we interrogate the oxidative stress responses of MEFs to the administration of three GOs. The intracellular ROS levels in MEFs

[(Fig._1)TD$IG]

Fig. 1. Characterization of the GOs with different oxidation degrees. (A) AFM images and size distributions of GO. (B) High-resolution C1s XPS scans of GO-h, GO-m and GO-l film on Si substrate. (C) Raman spectra of the GO samples.

W. Zhang et al. / Toxicology Letters 237 (2015) 61–71

[(Fig._2)TD$IG]

65

Fig. 2. Morphology observations of MEFs treated with GO-h, GO-m and GO-l at 50 mg/ml for 24 h. (A) TEM observations of the GO aggregates taken up by MEFs. The images on the bottom are high-magnification images of the boxed-in photos on top. The white and black arrows indicate GO aggregates inside and outside cells, respectively. (B) Optical photos of MEF cells treated with the GOs at 50 mg/ml for 24 h taken by Cytell Imaging System. Scale bars, 50 mm.

with 24-h GO pre-treatment were tested using H2DCF-DA (Kampkotter et al., 2007). As seen in Fig. 4, our results show that mean fluorescent intensities of MEFs treated with GO-h, GO-m and GO-l were all significantly higher than control, that is, 215%, 278% and 592% of that of control, respectively. Obviously, three GO samples induced dramatic enhancement in ROS levels in MEFs, and the decreased oxidation degree of GO resulted in higher ROS production. The hierarchical oxidative stress model suggests that high levels of ROS or oxidative stress in cells eventually result in cellular apoptosis or necrosis when the cellular defense system is overwhelmed (Nel et al., 2006). Our results suggest that the substantial oxidative stress induced by GO may be an important mechanism of the oxidation-degree dependent toxicity of the GOs on MEFs.

3.5. ESR experiments To further unravel the redox behaviors of the GOs with different oxidation degrees, we detected the ROS production by the GOs using electron spin resonance (ESR) spectrometry under in vitro chemical conditions. First of all, characteristic single peak spectra were presented for all GO samples (0.1 mg/ml) as shown in Fig. 5A. These one-peak spectra were measured to exhibit a g factor of 2.0023, indicating direct observation of unpaired electrons on GO samples (Prakash et al., 2013). The signal intensity increased with decreasing oxidation level, which is in line with the increasingly larger aromatic domain on nanosheets. ROS are by-products of aerobic metabolism, including superoxide anion (O2
–), hydrogen peroxide (H2O2) and hydroxyl radicals (
OH). Among them, H2O2 is much more stable than the

66

[(Fig._3)TD$IG]

W. Zhang et al. / Toxicology Letters 237 (2015) 61–71

Fig. 3. MEFs viability and apoptosis after 24-h treatment of GO-h, GO-m and GO-l. (A) MEFs viability measured by Cytell Cell Imaging System. *p < 0.05; **p < 0.01. (B) Comparison of cell percentage in different stages in apoptosis assays measured by flow cytometry after 24-h treatment of the GOs (50 mg/ml). The assays were conducted in triplicate. (C) Double staining analysis of MEF cells treated with or without the GOs (50 mg/ml) in apoptosis assays.

others and freely diffusible (Finkel and Holbrook, 2000). It is one of the most important signaling molecules in redox biology (Finkel, 2011), where it is involved in the oxidation of cysteine residues

within proteins (Rhee, 2006). Hence, we established a simple model of H2O2 decomposition by the ESR method to explore the possible chemical role of the three GOs on ROS conversion

W. Zhang et al. / Toxicology Letters 237 (2015) 61–71

(Gutierrez-Zepeda et al., 2005). Using BMPO as a spin trap (Abbas et al., 2014), no apparent ESR signal was detected either in the absence of either H2O2 or the GO samples (Fig. 5B, spectrum 1). However, when mixing H2O2 and the GOs, a 4-line spectrum with an intensity ratio of 1:2:2:1 was observed, which is attributable to the BMPO/
OH spin adduct (Fig. 5B, spectra 2–4). The less oxidized GO (GO-l) dramatically facilitated the
OH production from H2O2 decomposition. In addition to studying their influences on ROS conversion, the direct oxidative ability of the nanosheets were also detected. CPH can be used to determine the overall oxidative behavior of biological/chemical systems caused by holes and ROS (Weissmann et al., 2005). In Fig. 5C, our results showed that GO-l produced the highest signal intensity of CP
radicals, suggesting that GO-l possesses the strongest oxidative ability among three samples. 3.6. Theoretical simulation In order to illustrate the mechanisms why GO-l simulates the most
OH generation among three GO samples as in ESR experiments, we further investigated the underlying chemical mechanisms through theoretical simulation. As is shown in Fig. 6A, the carboxyl group of GO is initially oxidized into peroxycarboxyl group, and the energy barriers are independent on the size of aromatic domain. Then, the OO bond of the formed peroxycarboxyl homolytically cleaves, generating
OH and deprotonated carboxyl functionalized nanocarbon oxides, and the reaction energies of this step are intimately related to the size of aromatic domain. In particular, when the aromatic domain is 1 (Fig. 6B), the reaction energy as high as 39.4 kcal/mol; however, the reaction energy is only 23.4 kcal/mol, when the aromatic domain is 3. These results indicate that only those carboxyl functionalized graphene whose aromatic domain is large enough possesses ability to assist decomposition of H2O2, and this should stem from that large aromatic domain (as shown in Fig. 5A) can stabilize free electron of the formed deprotonated carboxyl functionalized GO.

67

4. Discussion Although impressive from a physicochemical viewpoint, the novel properties of graphene and GO raise concerns about their adverse effects on biological systems. However, the associations between the cytotoxicity and physicochemical properties of GO (particularly the oxygenated functional groups) and the underlying chemical mechanisms have not been explored in depth. In this work we synthesized three GO samples with various oxidation degrees by strictly controlling the amount of KMnO4 addition during synthesis process. The resulting GO-h, GO-m and GO-l all showed similar size distribution and good solubility in water. In this work we found the GO samples with C/O ratio higher than 3.0 (C/O ratio of GO-l is 2.9) could hardly be dispersed even in water. The GO aggregates into visible black particles that settles immediately after shaking and the liquid becomes clear (data not shown). Since our objective of the work is to study the impact of merely oxidation degree of nanosheets on their cytotoxicity, we tried to minimize the other physicochemical factors (e.g. size or solubility) that may interfere with experimental results. Therefore, we did not further investigate the toxicity of the GO with oxidation degree lower than GO-l. First of all, we confirmed that three GO samples could be taken up by MEFs. In our work GO was found in endosomes and/or lysosomes via TEM, both of whom are recognized as the most common localization compartments for nanomaterials (Chithrani and Chan, 2007; Peckys and de Jonge, 2011). Sasidharan et al. (2011) reported that carboxyl functionalized graphene could be internalized by monkey kidney cells, while pristine graphene accumulated on the cell membrane instead of being internalized by the cells. The internalization of nanosheets may be influenced by the properties of materials (e.g. size, dispersion and functionalized groups) and cell models. Morphological results show that, compared with GO-h and m, there were a larger number of cells in GO-l treated MEFs that displayed irregular shapes and smaller sizes with a large amount of cell fragments and round cells. Then the viability of MEFs treated

[(Fig._4)TD$IG]

Fig. 4. Intracellular ROS production in MEFs measured by flow cytometry. (A) H2DCF-DA staining analysis of MEF cells treated with or without the GOs (50 mg/ml). (B) Comparison of the mean fluorescent intensity between different groups. **p < 0.01.

68

[(Fig._5)TD$IG]

W. Zhang et al. / Toxicology Letters 237 (2015) 61–71

Fig. 5. ESR spectra of GO alone or GO interaction with H2O2. (A) ESR spectra of unpaired electron on GO-h, GO-m and GO-l (0.1 mg/ml) samples alone, respectively. (B) Demonstration of
OH generated by GO (0.2 mg/ml) in presence of H2O2 together with spin trap BMPO. Note that the “control (Ctrl) group” shown in the first line represents both control groups of “BMPO + H2O2” and “BMPO + GO-x” (x stands for h, m or l, respectively). (C) Oxidative status of GO (0.2 mg/ml) in 100 mM PBS buffer containing 0.1 mM CPH. The ESR spectra were recorded at 2 min after mixing the samples using 20 mW microwave power, 1 G field modulation, and 100 G scan width.

[(Fig._6)TD$IG]

Fig. 6. Theoretical simulation for the contributions of carboxyl groups and aromatic domain size to the reaction energy barrier. (A) Mechanism of GO assisting dissociation of H2O2. (B) Models used in the present work.

with the GOs was studied. Wang et al. (2011) and Ding et al. (2014) previously found that GO was not toxic to human fibroblast cells and T lymphocytes until the concentration rised up to 50 mg/ml and higher, respectively. In accordance with these, our work reveals that MEFs exhibit pretty good tolerance to the GO samples at lower dosages (25 and 12.5 mg/ml) but notable lower viability at higher concentrations (50 and 100 mg/ml). The consistency in the dose range of toxicity in these investigations suggests that 50 mg/ ml may be a threshold that is adequate for GO to exhibit toxicity on normal mammalian cells. It is notable that the heavy metal impurity in carbon nanomaterials is one of the most important factors for their toxicity (Ma et al., 2015). The only possible heavy metal impurities in our GO samples are Mn2+ ions that derived from KMnO4 oxidation and H2O2 reduction process. The content of Mn element in GO-l, GO-m and GO-h solution (50 mg/ml) is 0.347, 0.78 and 2.421 mg/ml, respectively (Fig. S5) as measured by ICP-MS

after acid digestion, four orders of magnitude lower than the reported toxic dose (250 mM) in neuronal cells, one of the most vulnerable cell types (Maddirala et al., 2015). Mn2+ content increases as the oxidation degree of GOs ascends, which is opposite to the toxicity tendency of three GO samples based on the assumption that Mn2+ impurities were toxic to MEFs. Therefore, we think that the cytotoxicity induced by Mn2+ impurities in GO samples can be negligible. The apoptosis results display that with the reduction of degrees of oxidation, the proportion of apoptosis rises while that of normal status reduces. The apoptotic changes caused by GO nanosheets were reported by other researchers (Akhavan et al., 2012b; Liao et al., 2011; Chang et al., 2011). During the early process of apoptosis, cell will experience shrinkage and the cell outlines become irregular (Kerr et al., 1972; Elmore, 2007). Thus the smaller and irregular shaped cells (Fig. 2B) may be undergoing early apoptosis process. From the results of

W. Zhang et al. / Toxicology Letters 237 (2015) 61–71

morphology, viability and apoptosis, it can be concluded that the GO cytotoxicity follows the oxidation-degree dependent manner, where GO toxicity ascends as the oxidation degree descends. Oxidative stress has been proposed as one of the major toxic mechanisms of nanomaterials (Sanchez et al., 2011; Lewinski et al., 2008; Li et al., 2008), which is caused by raised ROS levels that can oxidize various biomolecules including DNA, lipids and proteins and eventually induce apoptosis or necrosis (Nel et al., 2006). Hence, we then studied the oxidative stress response of GO treated MEF cells. Our results show that GO-l stimulated the highest ROS production in MEFs, up to about 6 times of that of control. The hierarchical oxidative stress model suggests that a moderate amount of oxidative stress activates the antioxidant response and restores cellular redox homeostasis, whereas high levels eventually result in cellular apoptosis or necrosis when the cellular defense system is overwhelmed (Nel et al., 2006). Therefore, the dramatic enhancement of ROS level in cells induced by the GOs suggests that the apoptosis (Fig. 3B and C) may be attributed to the high level of oxidative stress induced by the three GOs in our system; more importantly, it also implies that oxidative stress should be a crucial toxicity mechanism of GO on MEFs, where the less oxidized GO could produce higher intracellular ROS levels. The previous publications regarding in vitro toxicity of graphene-based materials suggest that, similar to other carbon nanomaterials (Zhao et al., 2015a), physicochemical characteristics may play a critical role in their biological effects of this novel class of nanomaterials (Sasidharan et al., 2011; Akhavan et al., 2012b; Yue et al., 2012). It has been suggested that GO nanosheets may possess electron transfer activity (Liu et al., 2011; Zuo et al., 2009). In particular, the ultrahigh electron mobility of graphene and its unique surface properties such as one-atom thickness makes it a reactive ability of facilitating electron transfer among ROS and protein macromolecules (Zuo et al., 2009). Our previous work also demonstrated that GO/PLL-PEG derivatives could facilitate H2O2 decomposition into
OH under chemical environment (Zhang et al., 2012). To unveil the possible relevant physicochemical mechanisms of their oxidation-degree dependent toxicity, we thus detected the effects of the GOs on ROS conversion in chemical conditions via ESR method. First of all, the one-peak spectra (Fig. 5A) present the direct evidence of unpaired electrons on GO samples (Prakash et al., 2013). The signal intensity increased with the decreasing oxidation level, which can be attributed to the increasingly larger aromatic domain on nanosheets (as shown in Fig. 5A). Secondly, using BMPO as spin trap, we found that GOs had significant facilitative effect on H2O2 decomposition into
OH. Under physiological condition, intracellular O2
– is primarily produced from the oxidation reactions by inherent oxidase enzymes (i.e. NOXs) or electron leakage from mitochondria. O2
– can be rapidly converted into H2O2 by superoxide dismutases (SODs). Then H2O2 will be converted to non-toxic H2O by cellular antioxidant proteins (i.e. peroxiredoxins, glutathione peroxidase and catalase). The third type of ROS,
OH is generated from H2O2 in presence of ferrous ions (i.e. Fenton reaction). Among various types of ROS, O2
– and H2O2 are involved in both redox signaling pathways and oxidative stress and do not indiscriminately damage proteins, whereas
OH is an extremely reactive free radical and indiscriminately induces oxidative damages (Schieber and Chandel, 2014). Therefore, cells have multiple mechanisms to maintain iron homeostasis to prevent the formation of toxic
OH (Schieber and Chandel, 2014). However, when the GOs are introduced into MEFs in our system, the H2O2 decomposition will be largely accelerated and
OH will accumulate within shorter time, thus breaking the balance of redox system and resulting in irreversible damage on cellular macromolecules, including lipids, proteins and DNA

69

(Dizdaroglu and Jaruga, 2012). The higher facilitative effect the GO exerts on H2O2 decomposition into
OH, the stronger oxidative damages the cells will be subject to. Thirdly, we also studied the direct oxidative ability of the GOs using CPH. CPH, though ESR silent itself, can be oxidized to form CP-nitroxide radicals (CP
) with a typical ESR spectrum of three lines with intensity ratios of 1:1:1 (Weissmann et al., 2005). CPH is not a specific spin label since it may react with a number of oxidants such as ROS and holes generated as charge carriers, and can be very slowly oxidized to form nitroxides by dissolved oxygen or irradiation. Therefore, CPH can be used to determine the overall oxidative behavior of biological/chemical systems caused by holes and ROS. Thus our results indicate that GO-l possesses the strongest oxidative ability among three samples. Taken together, our ESR results indicate that as the oxidation degree decreases: (1) the GO possesses more unpaired electron; (2) exerts stronger indirect oxidative damage on MEFs through facilitating
OH production from H2O2 more efficiently; (3) and imposes heavier oxidative burden on the cells through direct oxidative reactions. Obviously, in our results, the ability of the GO samples to promote
OH formation (Fig. 5B) was intimately associated with their oxidative ability (Fig. 5C), which strongly supported our observations in cytotoxicity, apoptosis and intracellular ROS assays. Next, theoretical simulation was further conducted to investigate the chemical mechanisms for the facilitative ability of GO on
OH production. Our previous work revealed that the carboxyl group of nanocarbon oxides, such as fullerene oxide, carbon nanotube oxide and graphene oxide, possesses great ability to facilitate of H2O2 decomposition and
OH generation, and that this ability closely depends on the size of aromatic domain of nanocarbon oxides (Zhao et al., 2015c). The reaction pathway is illustrated in Fig. 6A. XPS, AFM, fourier transform infrared spectroscopy (FTIR) and solid state 13C NMR spectra have characterized that GOs are large amorphous materials with carboxyl, hydroxyl, epoxy and so on, and recent theoretical works demonstrated that these oxygen functionalities tend to distribute in close proximity, dividing the GO plane into aromatic domains with different sizes (Park et al., 2008; Erickson et al., 2010; Gao et al., 2009; Robinson et al., 2011). Accordingly, for GO samples with nearly the same sizes, the higher the oxidation degree is, the smaller the aromatic domain is, which has been confirmed by our ESR experiments as shown in Fig. 5A. So we think that the less oxidized GO, GO-l, should correspond to model 3 (Fig. 6B), and that sample GO-h should correspond to model 1. Therefore, our calculation results are well in line with ESR results, i.e., that GO sample with low oxidation degree (model 3) possesses greater ability to facilitate decomposition of H2O2. Carbon nanotube (CNT) is the closest nanomaterial to graphene nanosheets (Geim and Novoselov, 2007). The toxicity of CNTs is reported to be heavily influenced by their functionalization degree (Sayes et al., 2006; Zhao et al., 2015a), i.e. carboxylation of CNTs makes CNTs abundant in oxygen atoms, and decreases their toxicity (Sayes et al., 2006). Our results provide important clues to understand the mechanisms of the lower toxicity of CNTs or even other carbon nanomaterials with larger amount of oxygenated groups. Furthermore, this work also lays foundation for the research of carbon nanomaterials functionalization (i.e. using polymers) (Zhao et al., 2015b) and for the development of nanobased platform for their broader applications in biomedicine (i.e. early diagnosis of cancer) (Li, 2015). 5. Conclusions In this work, we synthesized three GO samples with different oxidation state, and the resulting GOs have similar size distribution (200–260 nm) and all exhibited good solubility in water. Three GOs

70

W. Zhang et al. / Toxicology Letters 237 (2015) 61–71

can all be internalized by MEFs. The morphology, viability and apoptosis results indicate that the GO with lower degree of oxidation displayed stronger toxicity on MEFs. Meanwhile, three GO samples induced dramatic enhancement of ROS levels in cells, where the less oxidized GO stimulated higher intracellular ROS production. These reveal that oxidative stress may serve as the major mechanism of GO toxicity on MEFs. To unravel the physicochemical mechanisms of the oxidation-degree dependent toxicity of the GOs, the ESR assays were conducted. ESR data showed that with the decrease of oxidation degrees, GO possessed more free electrons, stronger oxidative ability and higher facilitative ability on
OH production from H2O2 and electron transfer during the reactions. Oxidative stress is a well-recognized toxicological mechanism of various nanoparticles (Sanchez et al., 2011; Lewinski et al., 2008; Li et al., 2008). Considering the indiscriminate ability of
OH to induce oxidative damages on various biomolecules, accelerating
OH production more efficiently by the less oxidized GOs is doomed to impose stronger oxidative damages on MEFs. Further, theoretical simulations indicate the key contributions of the carboxyl groups on the GO with larger aromatic domain to lowering the energy barrier of H2O2 decomposition reaction, thus facilitating the reaction process. These results strongly supported our ESR observation. All together, we showed the strong association between oxidation state and toxicity of GO and proposed a systemic chemical mechanism of the oxidation-state dependent cytotoxicity of GO on MEFs. More importantly, this work also provides new clues to better control the preparing process of other carbon nanomaterials for their safe applications in biomedical fields. Conflict of interest The authors declare that there are no conflicts of interest. Acknowledgements The authors thank Hongda Guo for the assistance with TEM, Xiao Song for his guidance in flow cytometry-based apoptosis assays, and Ruifang Zhao, Drs. Hanqing Chen, Yi Yuan, Yiye Li and Yanping Ding for their kind suggestions for this work. This work was supported by MoST 973 (2012CB934004; 2011CB933400); National Distinguished Young Scientists program (31325010); the Key Research Program of the Chinese Academy of Sciences, Grant NO. KGZD-EW-T06; and was partially supported by a regulatory science grant under the FDA Nanotechnology CORES Program. This article is not an official U.S. FDA guidance or policy statement. No official support or endorsement by the U.S. FDA is intended or should be inferred. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j. toxlet.2015.05.021. References Abbas, K., Hardy, M., Poulhes, F., Karoui, H., Tordo, P., Ouari, O., Peyrot, F., 2014. Detection of superoxide production in stimulated and unstimulated living cells using new cyclic nitrone spin traps. Free Radic. Bio. Med. 71, 281–290. Akhavan, O., Ghaderi, E., 2010. Toxicity of graphene and graphene oxide nanowalls against bacteria. ACS Nano 4, 5731–5736. Akhavan, O., Ghaderi, E., Akhavan, A., 2012a. Size-dependent genotoxicity of grapheme nanoplatelets in human stem cells. Biomaterials 33, 8017–8025. Akhavan, O., Ghaderi, E., Rahighi, R., 2012b. Toward single-dna electrochemical biosensing by graphene nanowalls. ACS Nano 6, 2904–2916.

An, J., Gou, Y., Yang, C., Hu, F., Wang, C., 2013. Synthesis of a biocompatible gelatin functionalized graphene nanosheets and its application for drug delivery. Mater. Sci. Eng. C 33, 2827–2837. Chang, Y., Yang, S., Liu, J., Dong, E., Wang, Y., Cao, A., Liu, Y., Wang, H., 2011. In vitro toxicity evaluation of graphene oxide on a549 cells. Toxicol. Lett. 200, 201–210. Chithrani, B.D., Chan, W.C.W., 2007. Elucidating the mechanism of cellular uptake and removal of protein-coated gold nanoparticles of different sizes and shapes. Nano Lett. 7, 1542–1550. Das, S., Singh, S., Singh, V., Joung, D., Dowding, J.M., Reid, D., Anderson, J., Zhai, L., Khondaker, S.I., Self, W.T., Seal, S., 2013. Oxygenated functional group density on graphene oxide: its effect on cell toxicity. Part. Part. Syst. Charact. 30, 148–157. Ding, Z., Zhang, Z., Ma, H., Chen, Y., 2014. In vitro hemocompatibility and toxic mechanism of graphene oxide on human peripheral blood t lymphocytes and serum albumin. ACS Appl. Mater. Interfaces 6, 19797–19807. Dizdaroglu, M., Jaruga, P., 2012. Mechanisms of free radical-induced damage to dna. Free Radic. Res. 46, 382–419. Elmore, S., 2007. Apoptosis: a review of programmed cell death. Toxicol. Pathol. 35, 495–516. Erickson, K., Erni, R., Lee, Z., Alem, N., Gannett, W., Zettl, A., 2010. Determination of the local chemical structure of graphene oxide and reduced graphene oxide. Adv. Mater. 22, 4467–4472. Finkel, T., 2011. Signal transduction by reactive oxygen species. J. Cell Biol. 194, 7–15. Finkel, T., Holbrook, N.J., 2000. Oxidants, oxidative stress and the biology of ageing. Nature 408, 239–247. Frisch, M.J., Trucks, G.W., Schlegel, H.B., Scuseria, G.E., Robb, M.A., Cheeseman, J.R., Scalmani, G., Barone, V., Mennucci, B., Petersson, G.A., Nakatsuji, H., Caricato, M., Li, X., Hratchian, H.P., Izmaylov, A.F., Bloino, J., Zheng, G., Sonnenberg, J.L., Hada, M., Ehara, M., Toyota, K., Fukuda, R., Hasegawa, J., Ishida, M., Nakajima, T., Honda, Y., Kitao, O., Nakai, H., Vreven, T., Montgomery, J.A., Peralta, J.E., Ogliaro, F., Bearpark, M., Heyd, J.J., Brothers, E., Kudin, K.N., Staroverov, V.N., Keith, T., Kobayashi, R., Normand, J., Raghavachari, K., Rendell, A., Burant, J.C., Iyengar, S.S., Tomasi, J., Cossi, M., Rega, N., Millam, J.M., Klene, M., Knox, J.E., Cross, J.B., Bakken, V., Adamo, C., Jaramillo, J., Gomperts, R., Stratmann, R.E., Yazyev, O., Austin, A.J., Cammi, R., Pomelli, C., Ochterski, J.W., Martin, R.L., Morokuma, K., Zakrzewski, V.G., Voth, G.A., Salvador, P., Dannenberg, J.J., Dapprich, S., Daniels, A.D., Farkas, O., Foresman, J.B., Ortiz, J.V., Cioslowski, J., Fox, D.J., 2010. Gaussian 09, Revision C.01. Gaussian, Inc., Wallingford CT. Gao, W., Alemany, L.B., Ci, L.J., Ajayan, P.M., 2009. New insights into the structure and reduction of graphite oxide. Nat. Chem. 1, 403–408. Geim, A.K., Novoselov, K.S., 2007. The rise of graphene. Nat. Mater. 6, 183–191. Gutierrez-Zepeda, A., Santell, R., Wu, Z.X., Brown, M., Wu, Y.J., Khan, I., Link, C.D., Zhao, B., Luo, Y., 2005. Soy isoflavone glycitein protects against beta amyloidinduced toxicity and oxidative stress in transgenic caenorhabditis elegans. BMC Neurosci. 6, 54–62. Hummers, W.S., Offeman, R.E., 1958. Preparation of graphitic oxide. J. Am. Chem. Soc. 80, 1339. Kampkotter, A., Nkwonkam, C.G., Zurawski, R.F., Timpel, C., Chovolou, Y., Watjen, W., Kahl, R., 2007. Investigations of protective effects of the flavonoids quercetin and rutin on stress resistance in the model organism caenorhabditis elegans. Toxicology 234, 113–123. Kerr, J.F., Wyllie, A.H., Currie, A.R., 1972. Apoptosis: a basic biological phenomenon with wide-ranging implications in tissue kinetics. Br. J. Cancer 26, 239–257. Lammel, T., Boisseaux, P., Fernandez-Cruz, M.L., Navas, J.M., 2013. Internalization and cytotoxicity of graphene oxide and carboxyl graphene nanoplatelets in the human hepatocellular carcinoma cell line hep g2. Part. Fibre Toxicol. 10 (1), 27. Lewinski, N., Colvin, V., Drezek, R., 2008. Cytotoxicity of nanoparticles. Small 4, 26– 49. Li, J., 2015. Nanotechnology-based platform for early diagnosis of cancer. Sci. Bull. 60, 488–490. Li, N., Xia, T., Nel, A.E., 2008. The role of oxidative stress in ambient particulate matter-induced lung diseases and its implications in the toxicity of engineered nanoparticles. Free Radic. Bio. Med. 4, 1689–1699. Liao, K., Lin, Y., Macosko, C.W., Haynes, C.L., 2011. Cytotoxicity of graphene oxide and graphene in human erythrocytes and skin fibroblasts. ACS Appl. Mater. Interfaces 3, 2607–2615. Liu, Y., Yu, D.S. Zeng, C. Miao, Z.C. Dai, 2010. Biocompatible graphene oxide-based glucose biosensors. Langmuir 26, 6158–6160. Liu, S., Zeng, T.H., Hofmann, M., Burcombe, E., Wei, J., Jiang, R., Kong, J., Chen, Y., 2011. Antibacterial activity of graphite, graphite oxide, graphene oxide, and reduced graphene oxide: membrane and oxidative stress. ACS Nano 5, 6971–6980. Liu, S., Hu, M. Zeng, T.H. Wu, R. Jiang, R. Wei, J. Wang, L. Kong, J. Chen, 2012. Lateral dimension-dependent antibacterial activity of graphene oxide sheets. Langmuir 28, 12364–12372. Luo, C., Li, Y., Yang, L., Zheng, Y., Long, J., Jia, J., Xiao, S., Liu, J., 2014. Activation of erk and p53 regulates copper oxide nanoparticle-induced cytotoxicity in keratinocytes and fibroblasts. Int. J. Nanomed. 9, 4763–4772. Ma, X., Yang, S.T., Tang, H., Liu, Y.F., Wang, H.F., 2015. Competitive adsorption of heavy metal ions on carbon nanotubes and the desorption in simulated biofluids. J. Colloid Interface Sci. 448, 347–355. Maddirala, Y., Tobwala, S., Ercal, N., 2015. N-acetylcysteineamide protects against manganese-induced toxicity in shsy5y cell line. Brain Res. 1608, 157–166. Nel, A., Xia, T., Madler, L., Li, N., 2006. Toxic potential of materials at the nanolevel. Science 311, 622–627. Pan, Y., Sahoo, N.G., Li, L., 2012. The application of graphene oxide in drug delivery. Expert Opin. Drug Delivery 9, 1365–1366.

W. Zhang et al. / Toxicology Letters 237 (2015) 61–71 Park, S., An, Piner, J., Jung, R.D., Yang, I., Velamakanni, D., Nguyen, A., Ruoff, S.T., RS, 2008. Aqueous suspension and characterization of chemically modified graphene sheets. Chem. Mater. 20, 6592–6594. Peckys, D.B., de Jonge, N., 2011. Visualizing gold nanoparticle uptake in live cells with liquid scanning transmission electron microscopy. Nano Lett. 11, 1733– 1738. Prakash, A., Misra, S.K., Bahadur, D., 2013. The role of reduced graphene oxide capping on defect induced ferromagnetism of zno nanorods. Nanotechnology 24, 1–10. Rhee, S., 2006. Cell signaling. h2o2, a necessary evil for cell signaling. Science 312, 1882–1883. Robinson, J.T., Tabakman, S.M., Liang, Y., Wang, H., Casalongue, H.S., Vinh, D., Dai, H., 2011. Ultrasmall reduced graphene oxide with high near-infrared absorbance for photothermal therapy. J. Am. Chem. Soc. 133, 6825–6831. Ryoo, S.R., Kim, Y.K., Kim, M.H., Min, D.H., 2010. Behaviors of nih-3t3 fibroblasts on graphene/carbon nanotubes: proliferation, focal adhesion, and gene transfection studies. ACS Nano 4, 6587–6598. Sanchez, V.C., Jachak, A., Hurt, R.H., Kane, A.B., 2011. Biological interactions of graphene-family nanomaterials: an interdisciplinary review. Chem. Res. Toxicol. 25, 15–34. Sasidharan, A., Panchakarla, L.S., Chandran, P., Menon, D., Nair, S., Rao, C.N.R., Koyakutty, M., 2011. Differential nano-bio interactions and toxicity effects of pristine versus functionalized graphene. Nanoscale 3, 2461–2464. Sayes, C.M., Liang, F., Hudson, J.L., Mendez, J., Guo,W, Beach, J.M., Moore, V.C., Doyle, C.D., West, J.L., Billups, W.E., Ausman, K.D., Colvin, V.L., 2006. Functionalization density dependence of single-walled carbon nanotubes cytotoxicity in vitro. Toxicol. Lett. 161, 135–142. Schieber, M., Chandel, N.S., 2014. Ros function in redox signaling and oxidative stress. Curr. Biol. 24, 453–462. Sun, X., Liu, Z., Welsher, K., Robinson, J.T., Goodwin, A., Zaric, S., Dai, H., 2008. Nanographene oxide for cellular imaging and drug delivery. Nano Res. 1, 203–212. Wang, K., Ruan, J., Song, H., Zhang, J., Wo, Y., Guo, S., Cui, D., 2011. Biocompatibility of graphene oxide. Nanoscale Res. Lett. 6, 1–8.

71

Weissmann, N., Kuzkaya, N., Fuchs, B., Tiyerili, V., Schafer, R.U., Schutte, H., Ghofrani, H.A., Schermuly, R.T., Schudt, C., Sydykov, A., Egemnazarow, B., Seeger, W., Grimminger, F., 2005. Detection of reactive oxygen species in isolated, perfused lungs by electron spin resonance spectroscopy. Respir. Res. 6, 86–100. Wu, T., Gao, J., Xu, X., Wang, W., Gao, C., Qiu, H., 2013. A new rapid chemical route to prepare reduced graphene oxide using copper metal nanoparticles. Nanotechnology 24, 215604–215613. Yan, L., Chang, Y., Zhao, L., Gu, Z., Liu, X., Tian, G., Zhou, L., Ren, W., Jin, S., Yin, W., Chang, H., Xing, G., Gao, X., Zhao, Y., 2013. The use of polyethylenimine-modified graphene oxide as a nanocarrier for transferring hydrophobic nanocrystals into water to produce water-dispersible hybrids for use in drug delivery. Carbon 57, 120–129. Yue, H., Wei, W., Yue, Z., Wang, B., Luo, N., Gao, Y., Ma, D., Ma, G., Su, Z., 2012. The role of the lateral dimension of graphene oxide in the regulation of cellular responses. Biomaterials 33, 4013–4021. Zhang, W., Guo, Z., Huang, D., Liu, Z., Guo, X., Zhong, H., 2011. Synergistic effect of chemo-photothermal therapy using pegylated graphene oxide. Biomaterials 32, 8555–8561. Zhang, W., Wang, C., Li, Z., Lu, Z., Li, Y., Yin, J., Zhou, Y., Gao, X., Fang, Y., Nie, G., Zhao, Y., 2012. Unraveling stress-induced toxicity properties of graphene oxide and the underlying mechanism. Adv. Mater. 24, 5391–5397. Zhao, F., Meng, H., Yan, L., Wang, B., Zhao, Y., 2015a. Nanosurface chemistry and dose govern the bioaccumulation and toxicity of carbon nanotubes, metal nanomaterials and quantum dots in vivo. Sci. Bull. 60, 3–20. Zhao, R., Wang, H., Ji, T., Anderson, G., Nie, G., Zhao, Y., 2015b. Biodegradable cationic e-poly-L-lysine-conjugated polymeric nanoparticles as a new effective antibacterial agent. Sci. Bull. 60, 216–226. Zhao, R., Zhao, X., Gao, X.F., 2015c. Molecular-level insights into intrinsic peroxidase-like activity of nanocarbon oxides. Chem. Eur. J. 21, 960–964. Zuo, X., He, S., Li, D., Peng, C., Huang, Q., Song, S., Fan, C., 2009. Graphene oxidefacilitated electron transfer of metalloproteins at electrode surfaces. Langmuir 26, 1936–1939.